scripts/02_url_classify_uk.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predicting Not News\n",
    "\n",
    "We predict not news using URL structure.\n",
    "\n",
    "Bakshy et al. for instance use the following logic:\n",
    "\n",
    "\thard_lab    Indicates whether the url (below) matches the regex: 'politi|usnews|world|national|state|elect|vote|govern|campaign|war|polic|econ|unemploy|racis|energy|abortion|educa|healthcare|immigration'\n",
    "\tsoft_lab    Indicates whether the url matches the regex: 'sports|entertainment|arts|fashion|style|lifestyle|leisure|celeb|movie|music|gossip|food|travel|horoscope|weather|gadget'\n",
    "\n",
    "\n",
    "The scheme is based on U.S. data so not as useful for U.K. data. After carefully inspecting a random sample of thousand URLs, we came up with the following:\n",
    "    \n",
    "\thard_lab    Indicates whether the url (below) matches the regex: 'politi|world|national|state|elect|vote|govern|campaign|war|polic|econ|unemploy|energy|educa|healthcare|immigration'\n",
    "\tsoft_lab    Indicates whether the url matches the regex: 'sports|entertainment|arts|fashion|style|lifestyle|leisure|celeb|movie|music|gossip|food|travel|horoscope|weather|gadget'\n",
    "\n",
    "p.s. 'soft_lab' means not news.\n",
    "\n",
    "### Read in the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "pd.set_option('display.max_colwidth', 120)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>text</th>\n",
       "      <th>url</th>\n",
       "      <th>src_name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>highpressur day work drink pint locallybrew real ale mani peopl prefer way relax anaesthetist aldebi near beccl turn...</td>\n",
       "      <td>http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&amp;category=NEWS&amp;tBrand=bbjo...</td>\n",
       "      <td>beccles and bungay journal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>weather forecast remain extrem cold come day suffolk counti council extend usual custom servic helplin oper weekend ...</td>\n",
       "      <td>http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&amp;category=NEWS&amp;tBrand=bbjo...</td>\n",
       "      <td>beccles and bungay journal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>elderli vulner peopl live council hous waveney protect new crackdown antisoci behaviour waveney district council new...</td>\n",
       "      <td>http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&amp;category=NEWS&amp;tBrand=bbjo...</td>\n",
       "      <td>beccles and bungay journal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>villag award energysav project set spend reduc fuel bill whole commun wenhaston south halesworth name one greenest v...</td>\n",
       "      <td>http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&amp;category=NEWS&amp;tBrand=bbjo...</td>\n",
       "      <td>beccles and bungay journal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>tuesday saw first beccl v bungay golf match year produc ace beccl play home led ladi captain sam cooper partner club...</td>\n",
       "      <td>http://www.becclesandbungayjournal.co.uk/content/bbjournal/sport/story.aspx?brand=BBJOnline&amp;category=SPORT&amp;tBrand=bb...</td>\n",
       "      <td>beccles and bungay journal</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                      text  \\\n",
       "0  highpressur day work drink pint locallybrew real ale mani peopl prefer way relax anaesthetist aldebi near beccl turn...   \n",
       "1  weather forecast remain extrem cold come day suffolk counti council extend usual custom servic helplin oper weekend ...   \n",
       "2  elderli vulner peopl live council hous waveney protect new crackdown antisoci behaviour waveney district council new...   \n",
       "3  villag award energysav project set spend reduc fuel bill whole commun wenhaston south halesworth name one greenest v...   \n",
       "4  tuesday saw first beccl v bungay golf match year produc ace beccl play home led ladi captain sam cooper partner club...   \n",
       "\n",
       "                                                                                                                       url  \\\n",
       "0  http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&category=NEWS&tBrand=bbjo...   \n",
       "1  http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&category=NEWS&tBrand=bbjo...   \n",
       "2  http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&category=NEWS&tBrand=bbjo...   \n",
       "3  http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&category=NEWS&tBrand=bbjo...   \n",
       "4  http://www.becclesandbungayjournal.co.uk/content/bbjournal/sport/story.aspx?brand=BBJOnline&category=SPORT&tBrand=bb...   \n",
       "\n",
       "                     src_name  \n",
       "0  beccles and bungay journal  \n",
       "1  beccles and bungay journal  \n",
       "2  beccles and bungay journal  \n",
       "3  beccles and bungay journal  \n",
       "4  beccles and bungay journal  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('/opt/notebooks/not_news/tabs/uk_news_media_data_subset_by_label.csv.gz', usecols=['label', 'url', 'text'])\n",
    "df['src_name'] = df['label'].str.lower()\n",
    "del df['label']\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5392953"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Generate Labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "hard_lab = re.compile('(politi|world|national|uk-news|scottish-news|news-eu|state|local|elect|vote|govern|campaign|war|polic|econ|unemploy|energy|educa|healthcare|immigration)')\n",
    "soft_lab = re.compile('(sport|football|entertainment|culture|arts|fashion|style|lifestyle|life-style|leisure|celeb|movie|music|gossip|food|travel|horoscope|weather|gadget)')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create Training Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>text</th>\n",
       "      <th>url</th>\n",
       "      <th>src_name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1294994</th>\n",
       "      <td>man name local geraint evan appear aberdar magistr court accus murder father paul bennett father two stab death grou...</td>\n",
       "      <td>http://www.peterboroughet.co.uk/viewarticle.aspx?articleid=2126845&amp;sectionid=406</td>\n",
       "      <td>peterborough evening telegraph</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2453364</th>\n",
       "      <td>lead polic chief today condemn armchair perfect health safeti taliban prevent offic carri job say put live risk sir ...</td>\n",
       "      <td>http://www.guardian.co.uk/Politics/homeaffairs/story/0,,2206622,00.htm</td>\n",
       "      <td>the guardian</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>121854</th>\n",
       "      <td>west brom bolton took point live barclay premier leagu encount hawthorn wander went front midway second half thank s...</td>\n",
       "      <td>http://www.birminghammail.net/birmingham-sport/west-bromwich-albion-fc/west-bromwich-albion-fc-news/2010/10/02/west-...</td>\n",
       "      <td>birmingham mail</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1077388</th>\n",
       "      <td>east england ambul servic norwich offic hellesdon east england ambul servic norwich offic hellesdon solar panel inst...</td>\n",
       "      <td>http://www.edp24.co.uk/solar_panels_installed_at_norwich_ambulance_control_room_1_225106</td>\n",
       "      <td>eastern daily press</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2926234</th>\n",
       "      <td>wit say man went stab spree belgian nurseri face paint like joker two babi carer die attack ten children treat hospi...</td>\n",
       "      <td>http://www.belfasttelegraph.co.uk/breaking-news/world/europe/belgian-nursery-killer-painted-face-like-the-joker-1415...</td>\n",
       "      <td>belfast telegraph</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                            text  \\\n",
       "1294994  man name local geraint evan appear aberdar magistr court accus murder father paul bennett father two stab death grou...   \n",
       "2453364  lead polic chief today condemn armchair perfect health safeti taliban prevent offic carri job say put live risk sir ...   \n",
       "121854   west brom bolton took point live barclay premier leagu encount hawthorn wander went front midway second half thank s...   \n",
       "1077388  east england ambul servic norwich offic hellesdon east england ambul servic norwich offic hellesdon solar panel inst...   \n",
       "2926234  wit say man went stab spree belgian nurseri face paint like joker two babi carer die attack ten children treat hospi...   \n",
       "\n",
       "                                                                                                                             url  \\\n",
       "1294994                                         http://www.peterboroughet.co.uk/viewarticle.aspx?articleid=2126845&sectionid=406   \n",
       "2453364                                                   http://www.guardian.co.uk/Politics/homeaffairs/story/0,,2206622,00.htm   \n",
       "121854   http://www.birminghammail.net/birmingham-sport/west-bromwich-albion-fc/west-bromwich-albion-fc-news/2010/10/02/west-...   \n",
       "1077388                                 http://www.edp24.co.uk/solar_panels_installed_at_norwich_ambulance_control_room_1_225106   \n",
       "2926234  http://www.belfasttelegraph.co.uk/breaking-news/world/europe/belgian-nursery-killer-painted-face-like-the-joker-1415...   \n",
       "\n",
       "                               src_name  \n",
       "1294994  peterborough evening telegraph  \n",
       "2453364                    the guardian  \n",
       "121854                  birmingham mail  \n",
       "1077388             eastern daily press  \n",
       "2926234               belfast telegraph  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sdf = df[df.text.notnull() & df.url.notnull()].sample(1000000, random_state=4).copy()\n",
    "sdf.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Free up memory\n",
    "import gc\n",
    "%xdel df\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>text</th>\n",
       "      <th>url</th>\n",
       "      <th>src_name</th>\n",
       "      <th>hard_lab</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1294994</th>\n",
       "      <td>man name local geraint evan appear aberdar magistr court accus murder father paul bennett father two stab death grou...</td>\n",
       "      <td>http://www.peterboroughet.co.uk/viewarticle.aspx?articleid=2126845&amp;sectionid=406</td>\n",
       "      <td>peterborough evening telegraph</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2926234</th>\n",
       "      <td>wit say man went stab spree belgian nurseri face paint like joker two babi carer die attack ten children treat hospi...</td>\n",
       "      <td>http://www.belfasttelegraph.co.uk/breaking-news/world/europe/belgian-nursery-killer-painted-face-like-the-joker-1415...</td>\n",
       "      <td>belfast telegraph</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2950250</th>\n",
       "      <td>monday januari local news member ahmadiyya muslim youth associ member ahmadiyya muslim youth associ member ahmadiyya...</td>\n",
       "      <td>http://www.croydonguardian.co.uk/news/localnews/10175440.Muslim_group_pitches_in_to_clear_snow/?ref=rs</td>\n",
       "      <td>croydon guardian</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5005952</th>\n",
       "      <td>england face prospect follow india oval anoth lowerord bat collaps face task score final match seri ensur indian bat...</td>\n",
       "      <td>http://www.skegnesstoday.co.uk/latest-national-sport?articleid=310453</td>\n",
       "      <td>skegness today</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3827845</th>\n",
       "      <td>polic today reveal lead forc new neighbourhood team follow costcut shakeup commun beat dian bradburi number individu...</td>\n",
       "      <td>http://www.newsandstar.co.uk/news/cumbria-police-reveals-inspectors-in-charge-of-force-s-new-local-teams-1.799776?re...</td>\n",
       "      <td>news &amp; star</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                            text  \\\n",
       "1294994  man name local geraint evan appear aberdar magistr court accus murder father paul bennett father two stab death grou...   \n",
       "2926234  wit say man went stab spree belgian nurseri face paint like joker two babi carer die attack ten children treat hospi...   \n",
       "2950250  monday januari local news member ahmadiyya muslim youth associ member ahmadiyya muslim youth associ member ahmadiyya...   \n",
       "5005952  england face prospect follow india oval anoth lowerord bat collaps face task score final match seri ensur indian bat...   \n",
       "3827845  polic today reveal lead forc new neighbourhood team follow costcut shakeup commun beat dian bradburi number individu...   \n",
       "\n",
       "                                                                                                                             url  \\\n",
       "1294994                                         http://www.peterboroughet.co.uk/viewarticle.aspx?articleid=2126845&sectionid=406   \n",
       "2926234  http://www.belfasttelegraph.co.uk/breaking-news/world/europe/belgian-nursery-killer-painted-face-like-the-joker-1415...   \n",
       "2950250                   http://www.croydonguardian.co.uk/news/localnews/10175440.Muslim_group_pitches_in_to_clear_snow/?ref=rs   \n",
       "5005952                                                    http://www.skegnesstoday.co.uk/latest-national-sport?articleid=310453   \n",
       "3827845  http://www.newsandstar.co.uk/news/cumbria-police-reveals-inspectors-in-charge-of-force-s-new-local-teams-1.799776?re...   \n",
       "\n",
       "                               src_name  hard_lab  \n",
       "1294994  peterborough evening telegraph       1.0  \n",
       "2926234               belfast telegraph       1.0  \n",
       "2950250                croydon guardian       1.0  \n",
       "5005952                  skegness today       1.0  \n",
       "3827845                     news & star       1.0  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def is_hard_lab(c):\n",
    "    m = hard_lab.search(c)\n",
    "    return 1 if m else None\n",
    "\n",
    "sdf['hard_lab'] = sdf['url'].apply(lambda c: is_hard_lab(c))\n",
    "sdf[sdf.hard_lab == 1].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>text</th>\n",
       "      <th>url</th>\n",
       "      <th>src_name</th>\n",
       "      <th>hard_lab</th>\n",
       "      <th>soft_lab</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>121854</th>\n",
       "      <td>west brom bolton took point live barclay premier leagu encount hawthorn wander went front midway second half thank s...</td>\n",
       "      <td>http://www.birminghammail.net/birmingham-sport/west-bromwich-albion-fc/west-bromwich-albion-fc-news/2010/10/02/west-...</td>\n",
       "      <td>birmingham mail</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5005952</th>\n",
       "      <td>england face prospect follow india oval anoth lowerord bat collaps face task score final match seri ensur indian bat...</td>\n",
       "      <td>http://www.skegnesstoday.co.uk/latest-national-sport?articleid=310453</td>\n",
       "      <td>skegness today</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2184529</th>\n",
       "      <td>coast downhil gear cheaper honest john explain coast downhil gear cheaper fuel effici gearbox engag honest john bst ...</td>\n",
       "      <td>http://telegraph.feedsportal.com/c/32726/f/568313/s/161c1511/l/0L0Stelegraph0O0Cmotoring0Ccaradvice0Chonestjohn0C859...</td>\n",
       "      <td>the daily telegraph</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3160926</th>\n",
       "      <td>villa seal collin deal sep aston villa complet sign west ham defend jame collin fouryear contract welsh intern put p...</td>\n",
       "      <td>http://www.examiner.co.uk/news/national-sport/2009/09/01/villa-seal-collins-deal-86081-24582146</td>\n",
       "      <td>the huddersfield daily examiner</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5224091</th>\n",
       "      <td>late show delight lambert sunday januari nation sport news press associ skip social link print email tweet paul lamb...</td>\n",
       "      <td>http://www.somersetcountygazette.co.uk/uk_national_sport/9446365.Late_show_delights_Lambert/?ref=rs</td>\n",
       "      <td>somerset county gazette</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                            text  \\\n",
       "121854   west brom bolton took point live barclay premier leagu encount hawthorn wander went front midway second half thank s...   \n",
       "5005952  england face prospect follow india oval anoth lowerord bat collaps face task score final match seri ensur indian bat...   \n",
       "2184529  coast downhil gear cheaper honest john explain coast downhil gear cheaper fuel effici gearbox engag honest john bst ...   \n",
       "3160926  villa seal collin deal sep aston villa complet sign west ham defend jame collin fouryear contract welsh intern put p...   \n",
       "5224091  late show delight lambert sunday januari nation sport news press associ skip social link print email tweet paul lamb...   \n",
       "\n",
       "                                                                                                                             url  \\\n",
       "121854   http://www.birminghammail.net/birmingham-sport/west-bromwich-albion-fc/west-bromwich-albion-fc-news/2010/10/02/west-...   \n",
       "5005952                                                    http://www.skegnesstoday.co.uk/latest-national-sport?articleid=310453   \n",
       "2184529  http://telegraph.feedsportal.com/c/32726/f/568313/s/161c1511/l/0L0Stelegraph0O0Cmotoring0Ccaradvice0Chonestjohn0C859...   \n",
       "3160926                          http://www.examiner.co.uk/news/national-sport/2009/09/01/villa-seal-collins-deal-86081-24582146   \n",
       "5224091                      http://www.somersetcountygazette.co.uk/uk_national_sport/9446365.Late_show_delights_Lambert/?ref=rs   \n",
       "\n",
       "                                src_name  hard_lab  soft_lab  \n",
       "121854                   birmingham mail       NaN       1.0  \n",
       "5005952                   skegness today       1.0       1.0  \n",
       "2184529              the daily telegraph       NaN       1.0  \n",
       "3160926  the huddersfield daily examiner       1.0       1.0  \n",
       "5224091          somerset county gazette       1.0       1.0  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def is_soft_lab(c):\n",
    "    m = soft_lab.search(c)\n",
    "    return 1 if m else None\n",
    "\n",
    "sdf['soft_lab'] = sdf['url'].apply(lambda c: is_soft_lab(c))\n",
    "sdf[sdf.soft_lab == 1].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Filter on rows where hard and soft labels are different\n",
    "\n",
    "We ignore URLs that have both keywords."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "sdf.loc[(sdf.hard_lab == 1) & (sdf.soft_lab.isnull()), 'hard_soft'] = 1\n",
    "sdf.loc[(sdf.hard_lab.isnull()) & (sdf.soft_lab == 1), 'hard_soft'] = 0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Final Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>text</th>\n",
       "      <th>url</th>\n",
       "      <th>src_name</th>\n",
       "      <th>hard_lab</th>\n",
       "      <th>soft_lab</th>\n",
       "      <th>hard_soft</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>man name local geraint evan appear aberdar magistr court accus murder father paul bennett father two stab death grou...</td>\n",
       "      <td>http://www.peterboroughet.co.uk/viewarticle.aspx?articleid=2126845&amp;sectionid=406</td>\n",
       "      <td>peterborough evening telegraph</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>west brom bolton took point live barclay premier leagu encount hawthorn wander went front midway second half thank s...</td>\n",
       "      <td>http://www.birminghammail.net/birmingham-sport/west-bromwich-albion-fc/west-bromwich-albion-fc-news/2010/10/02/west-...</td>\n",
       "      <td>birmingham mail</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>wit say man went stab spree belgian nurseri face paint like joker two babi carer die attack ten children treat hospi...</td>\n",
       "      <td>http://www.belfasttelegraph.co.uk/breaking-news/world/europe/belgian-nursery-killer-painted-face-like-the-joker-1415...</td>\n",
       "      <td>belfast telegraph</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>monday januari local news member ahmadiyya muslim youth associ member ahmadiyya muslim youth associ member ahmadiyya...</td>\n",
       "      <td>http://www.croydonguardian.co.uk/news/localnews/10175440.Muslim_group_pitches_in_to_clear_snow/?ref=rs</td>\n",
       "      <td>croydon guardian</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>polic today reveal lead forc new neighbourhood team follow costcut shakeup commun beat dian bradburi number individu...</td>\n",
       "      <td>http://www.newsandstar.co.uk/news/cumbria-police-reveals-inspectors-in-charge-of-force-s-new-local-teams-1.799776?re...</td>\n",
       "      <td>news &amp; star</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                      text  \\\n",
       "0  man name local geraint evan appear aberdar magistr court accus murder father paul bennett father two stab death grou...   \n",
       "1  west brom bolton took point live barclay premier leagu encount hawthorn wander went front midway second half thank s...   \n",
       "2  wit say man went stab spree belgian nurseri face paint like joker two babi carer die attack ten children treat hospi...   \n",
       "3  monday januari local news member ahmadiyya muslim youth associ member ahmadiyya muslim youth associ member ahmadiyya...   \n",
       "4  polic today reveal lead forc new neighbourhood team follow costcut shakeup commun beat dian bradburi number individu...   \n",
       "\n",
       "                                                                                                                       url  \\\n",
       "0                                         http://www.peterboroughet.co.uk/viewarticle.aspx?articleid=2126845&sectionid=406   \n",
       "1  http://www.birminghammail.net/birmingham-sport/west-bromwich-albion-fc/west-bromwich-albion-fc-news/2010/10/02/west-...   \n",
       "2  http://www.belfasttelegraph.co.uk/breaking-news/world/europe/belgian-nursery-killer-painted-face-like-the-joker-1415...   \n",
       "3                   http://www.croydonguardian.co.uk/news/localnews/10175440.Muslim_group_pitches_in_to_clear_snow/?ref=rs   \n",
       "4  http://www.newsandstar.co.uk/news/cumbria-police-reveals-inspectors-in-charge-of-force-s-new-local-teams-1.799776?re...   \n",
       "\n",
       "                         src_name  hard_lab  soft_lab  hard_soft  \n",
       "0  peterborough evening telegraph       1.0       NaN        1.0  \n",
       "1                 birmingham mail       NaN       1.0        0.0  \n",
       "2               belfast telegraph       1.0       NaN        1.0  \n",
       "3                croydon guardian       1.0       NaN        1.0  \n",
       "4                     news & star       1.0       NaN        1.0  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sdf = sdf[sdf.hard_soft.notnull()]\n",
    "sdf = sdf.reset_index(drop=True)\n",
    "sdf.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "372163"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sdf.shape[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Modeling\n",
    "\n",
    "Split into train and test, and get non-sparse and moderately frequent bi- and tri-grams."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "train_df, test_df = train_test_split(sdf, test_size=0.2, random_state=4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Build the bi- and tri-grams that appear in more than 50 outlets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<266x46711 sparse matrix of type '<class 'numpy.int64'>'\n",
       "\twith 3459361 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "\n",
    "vectorizer = CountVectorizer(\n",
    "  min_df=50,\n",
    "  ngram_range = (2, 3),\n",
    "  binary=True,\n",
    "  stop_words='english')\n",
    "vectorizer.fit_transform(sdf.groupby('src_name')['text'].agg(lambda x:' '.join(x)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "46711"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vocab_set = set(vectorizer.vocabulary_)\n",
    "len(vocab_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting features from the training dataset using a sparse vectorizer\n",
      "done in 1494.385205s at 0.340MB/s\n",
      "n_samples: 297730, n_features: 60323\n"
     ]
    }
   ],
   "source": [
    "# Now classify w sklearn:\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "from sklearn import metrics\n",
    "import numpy as np\n",
    "from time import time\n",
    "\n",
    "# Create text features (and store efficiently as sparse matrix).\n",
    "print(\"Extracting features from the training dataset using a sparse vectorizer\")\n",
    "t0 = time()\n",
    "\n",
    "vectorizer = TfidfVectorizer(\n",
    "  sublinear_tf=False,\n",
    "  max_df=0.5,\n",
    "  min_df=100,\n",
    "  ngram_range = (2,3),\n",
    "  use_idf=True,\n",
    "  max_features=500000,\n",
    "  stop_words='english')\n",
    "\n",
    "def size_mb(docs):\n",
    "      return sum(len(s.encode('utf-8')) for s in docs) / 1e6\n",
    "\n",
    "t0 = time()\n",
    "data_train_size_mb = size_mb(train_df['text'])\n",
    "X_traintfidf = vectorizer.fit_transform(train_df['text'])\n",
    "duration = time() - t0\n",
    "print(\"done in %fs at %0.3fMB/s\" % (duration, data_train_size_mb / duration))\n",
    "print(\"n_samples: %d, n_features: %d\" % X_traintfidf.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "vocab = []\n",
    "for v in list(vectorizer.vocabulary_):\n",
    "    if v in vocab_set:\n",
    "        vocab.append(v)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "37582"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(vocab)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "done in 223.781821s at 2.272MB/s\n",
      "n_samples: 297730, n_features: 37582\n"
     ]
    }
   ],
   "source": [
    "vectorizer = TfidfVectorizer(\n",
    "  sublinear_tf=False,\n",
    "  ngram_range = (2,3),\n",
    "  use_idf=True,\n",
    "  vocabulary=vocab,\n",
    "  stop_words='english')\n",
    "\n",
    "t0 = time()\n",
    "data_train_size_mb = size_mb(train_df['text'])\n",
    "X_traintfidf = vectorizer.fit_transform(train_df['text'])\n",
    "duration = time() - t0\n",
    "print(\"done in %fs at %0.3fMB/s\" % (duration, data_train_size_mb / duration))\n",
    "print(\"n_samples: %d, n_features: %d\" % X_traintfidf.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Output Random Sample of Train and Test\n",
    "\n",
    "This is to estimate generalization error and also get a sense of how often we have bad labels in the training data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df.sample(1000, random_state=4).to_csv('sample/train_sample.csv', index_label='id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_df.sample(1000, random_state=4).to_csv('sample/test_sample.csv', index_label='id')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Regularized Logistic Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegressionCV(Cs=10, class_weight=None, cv=5, dual=False,\n",
       "           fit_intercept=True, intercept_scaling=1.0, max_iter=300,\n",
       "           multi_class='warn', n_jobs=-1, penalty='l2', random_state=42,\n",
       "           refit=True, scoring=None, solver='lbfgs', tol=0.0001, verbose=0)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.linear_model import LogisticRegressionCV\n",
    "\n",
    "# Now fit the model:\n",
    "clf = LogisticRegressionCV(solver='lbfgs', cv=5, n_jobs=-1, random_state=42, max_iter=300)\n",
    "clf.fit(X_traintfidf, train_df['hard_soft'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([0.35938137]),\n",
       " array([[[  5,   7,  13,  23,  60, 179, 300, 300, 300, 300],\n",
       "         [  5,   7,  13,  22,  71, 288, 300, 300, 300, 300],\n",
       "         [  5,   7,  13,  23,  47, 258, 300, 300, 300, 300],\n",
       "         [  5,   7,  13,  24,  80, 192, 300, 300, 300, 300],\n",
       "         [  5,   7,  13,  24,  59, 143, 300, 300, 300, 300]]], dtype=int32),\n",
       " array([1.00000000e-04, 7.74263683e-04, 5.99484250e-03, 4.64158883e-02,\n",
       "        3.59381366e-01, 2.78255940e+00, 2.15443469e+01, 1.66810054e+02,\n",
       "        1.29154967e+03, 1.00000000e+04]))"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.C_, clf.n_iter_, clf.Cs_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Most informative features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t-19.6812\trelat articl   \t\t10.6918\tnation news    \n",
      "\t-18.2190\temail text     \t\t6.8437\tcomment new    \n",
      "\t-5.6515\tlarg section   \t\t6.2165\treport abus    \n",
      "\t-5.3597\tlaw order      \t\t4.8785\tlocal news     \n",
      "\t-4.4596\tpremier leagu  \t\t4.7825\tgordon brown   \n",
      "\t-4.0331\temail friend print\t\t4.5157\tcomment articl \n",
      "\t-3.8741\tfriend print page\t\t4.3036\tprime minist   \n",
      "\t-3.8685\tend season     \t\t4.1943\tpolic offic    \n",
      "\t-3.8243\tfriend print   \t\t3.8719\tbarack obama   \n",
      "\t-3.8042\tlive council   \t\t3.7052\tcomment post   \n",
      "\t-3.7789\tchampion leagu \t\t3.4138\tperson use     \n",
      "\t-3.7720\ttv news        \t\t3.3575\tkill peopl     \n",
      "\t-3.3591\tmanchest citi  \t\t3.3364\tpolic said     \n",
      "\t-3.3487\tsecond half    \t\t3.2830\tgener elect    \n",
      "\t-3.2939\taston villa    \t\t3.2808\tquot http      \n",
      "\t-3.2815\tcelebr year    \t\t3.2470\tpolic station  \n",
      "\t-3.1895\tgrand prix     \t\t3.2131\tpeopl kill     \n",
      "\t-3.1670\tbox offic      \t\t3.1572\tcomment http   \n",
      "\t-3.0891\tfood safeti    \t\t3.1271\tforeign minist \n",
      "\t-3.0801\ttravel site    \t\t3.1199\toffici said    \n",
      "\t-3.0187\tdebt crisi     \t\t3.1052\thigher educ    \n",
      "\t-2.9786\taccrington stanley\t\t3.0847\tpress associ   \n",
      "\t-2.9683\tblackburn rover\t\t3.0756\tliber democrat \n",
      "\t-2.9246\twest ham       \t\t3.0484\tanyon inform   \n",
      "\t-2.8561\tlondon olymp   \t\t3.0180\tconsid action  \n",
      "\t-2.8241\tbritain apo    \t\t3.0098\tnews stori     \n",
      "\t-2.8169\tnew album      \t\t2.9534\tunit kingdom   \n",
      "\t-2.7765\tcounti championship\t\t2.9500\tlocal govern   \n",
      "\t-2.7633\theineken cup   \t\t2.8594\tmedia group    \n",
      "\t-2.7397\toliv oil       \t\t2.8111\thillari clinton\n",
      "\t-2.7368\tfootbal associ \t\t2.7942\tadd comment    \n",
      "\t-2.6953\toxford unit    \t\t2.7612\tshot dead      \n",
      "\t-2.6917\tlatest news    \t\t2.7339\tscottish govern\n",
      "\t-2.6787\tleisur centr   \t\t2.6582\tcrown court    \n",
      "\t-2.6646\tfood drink     \t\t2.6506\tspokeswoman said\n",
      "\t-2.6636\tstart season   \t\t2.6362\twar crime      \n",
      "\t-2.6619\talex mcleish   \t\t2.6088\tend quot       \n",
      "\t-2.6545\tleagu divis    \t\t2.5876\treport said    \n",
      "\t-2.6446\ttv radio       \t\t2.5507\treport problem \n",
      "\t-2.6404\twest brom      \t\t2.5151\tappear press   \n",
      "\t-2.6223\treal madrid    \t\t2.4750\tview express   \n",
      "\t-2.5794\toffic uk       \t\t2.4684\tpresidenti elect\n",
      "\t-2.5635\thead coach     \t\t2.4655\tpolic investig \n",
      "\t-2.5512\tautumn statement\t\t2.4646\tmr brown       \n",
      "\t-2.5203\tdaili telegraph\t\t2.4602\tmedia report   \n",
      "\t-2.5191\tnew manag      \t\t2.4466\tnews agenc     \n",
      "\t-2.5101\ttransfer window\t\t2.4354\tministri defenc\n",
      "\t-2.5059\tipswich town   \t\t2.4309\tsaid mr        \n",
      "\t-2.5047\tknee injuri    \t\t2.4227\tnation award   \n",
      "\t-2.4969\tsuper leagu    \t\t2.4213\tbillion euro   \n"
     ]
    }
   ],
   "source": [
    "def show_most_informative_features(vectorizer, clf, n=20):\n",
    "    c_f = sorted(zip(clf.coef_[0], vectorizer.get_feature_names()))\n",
    "    top = zip(c_f[:n], c_f[:-(n+1):-1])\n",
    "    for (c1,f1),(c2,f2) in top:\n",
    "        print(\"\\t%.4f\\t%-15s\\t\\t%.4f\\t%-15s\" % (c1,f1,c2,f2))\n",
    "\n",
    "show_most_informative_features(vectorizer=vectorizer, clf=clf, n=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_testtfidf = vectorizer.transform(test_df['text'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<74433x37582 sparse matrix of type '<class 'numpy.float64'>'\n",
       "\twith 2779732 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_testtfidf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Performance in Test Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0       0.89      0.85      0.87     35912\n",
      "         1.0       0.86      0.90      0.88     38521\n",
      "\n",
      "   micro avg       0.88      0.88      0.88     74433\n",
      "   macro avg       0.88      0.87      0.88     74433\n",
      "weighted avg       0.88      0.88      0.88     74433\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Predict on test data:\n",
    "y_test_pred = clf.predict(X_testtfidf)\n",
    "print(metrics.classification_report(test_df['hard_soft'], y_test_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Performance in Training Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0       0.90      0.86      0.88    143730\n",
      "         1.0       0.87      0.91      0.89    154000\n",
      "\n",
      "   micro avg       0.88      0.88      0.88    297730\n",
      "   macro avg       0.89      0.88      0.88    297730\n",
      "weighted avg       0.89      0.88      0.88    297730\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_train_pred = clf.predict(X_traintfidf)\n",
    "print(metrics.classification_report(train_df['hard_soft'], y_train_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Hard News"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index</th>\n",
       "      <th>text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2457</th>\n",
       "      <td>4536</td>\n",
       "      <td>everyon knew week would special one life sport bond guarante jackpot offer member draw eagerli anticip time appreci memor week incred jackpot winner found first draw number winner noel aiken member boyd dougla agenc genial noel member sinc week one sport bond nine year ago former gallah employe ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37084</th>\n",
       "      <td>68626</td>\n",
       "      <td>add stori itv sport involv major polit confront tv giant al jazeera could cost english network multimillion pound champion leagu product contract middl east al jazeera present gari linek arab station bbc gari linek present widespread coverag competit new sport channel itv sport programm product ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14392</th>\n",
       "      <td>26626</td>\n",
       "      <td>david suchet apo poirot final board orient express year play sleuth david suchet tell naomi west version excit disturb david suchet star agatha christi apo poirot murder orient express christma day david suchet star agatha christi apo poirot murder orient express christma day photo itv naomi wes...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21087</th>\n",
       "      <td>39028</td>\n",
       "      <td>come plantat peac question pose new north west busi leader chamber agm last week jim sammon join rank host high profil busi champion led chamber commerc movement nw incom presid jim laid key prioriti two year term offic clear whilst derri region capit citi therefor chamber region leader new pres...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33050</th>\n",
       "      <td>61161</td>\n",
       "      <td>teenag row die drag yard polic station stab wound daili mail report last updat pm septemb add stori victim craig marshal teenag die result violenc london year teenag die walk polic station blood pour stab wound craig marshal knife repeatedli row manag stagger yard acton polic station west london...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39541</th>\n",
       "      <td>73204</td>\n",
       "      <td>monday decemb oxford utd oxford mail photograph author david pritchard chief sport report cover oxford unit follow us twitter oxfordmailoufc call oxford mail rodney parad play surfac morn pictur mark jone rodney parad play surfac morn pictur mark jone wet weather wale cast doubt oxford unit visi...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18807</th>\n",
       "      <td>34808</td>\n",
       "      <td>polic offic accus collud colleagu arrest young protest fals ground student fee demonstr polic watchdog said investig seriou alleg offic abus posit breach cordon highprofil demonstr parliament squar last decemb rachel cerfontyn commission independ polic complaint commiss said also look circumst m...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25285</th>\n",
       "      <td>46753</td>\n",
       "      <td>vladimir putin ever take anyth worri replac someth bigger better address alleg stole ring billionair robert kraft award american footbal team own super bowl russian presid said know rememb neither mr kraft ring valuabl thing kraft team propos putin said respons question annual econom forum st pe...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4557</th>\n",
       "      <td>8477</td>\n",
       "      <td>jenni perciv westminst editor gordon brown scottish mp exclud vote englandonli law radic conserv plan unveil week tori leader david cameron pois back major overhaul westminst system english mp would meet separ grand committe decid legisl scotland mp would expect sit full vote common would speake...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30239</th>\n",
       "      <td>55940</td>\n",
       "      <td>rsa insur rose tumultu insur sector activist investor appear compani sharehold regist share beleagu group tuesday start deepli discount right issu advanc cevian capit disclos stake compani swedish investor selfdescrib activist success push demerg engin cookson previous agit chang reinsur munich ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15881</th>\n",
       "      <td>29377</td>\n",
       "      <td>david cameron travel brussel eu summit help endors place munich articl time professor econom univers munich hanswern sinn prais prime minist recent speech eu essenc mr cameron right someth amiss eu regul far mani thing fall outsid remit prime minist new fan germani appear may struggl convinc adv...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14662</th>\n",
       "      <td>27150</td>\n",
       "      <td>secondjob offic manag owe tax may liz keen western mail compani hous manag took second job make end meet given suspend jail sentenc admit evad incom tax john llewellyn end owe inland revenu seven year parttim deliveri driver wh smith cardiff crown court heard start retail earn inform technolog m...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31811</th>\n",
       "      <td>58837</td>\n",
       "      <td>publish briton ever christma shop onlin decemb almost one five nonfood purchas made internet web sale growth acceler compar month fastest rate three year overal uk retail sale grew likeforlik basi data british retail consortium brc survey carri kpmg show onlin trade repres total nonfood sale dec...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12307</th>\n",
       "      <td>22722</td>\n",
       "      <td>man charg connect threehour bank sieg polic said around peopl taken hostag monday barclay bank church road ashford west london matthew nutley brooksid avenu ashford charg robberi fals imprison possess imit firearm intent commit robberi two count make threat kill also charg blackmail possess offe...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7835</th>\n",
       "      <td>14483</td>\n",
       "      <td>investig claim doctor grant women illeg abort base sex unborn babi launch depart health spark undercov newspap investig sexselect abort secretli film doctor british clinic agre termin foetus either male femal doctor also allegedli record admit prepar falsifi paperwork arrang illeg abort health s...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20950</th>\n",
       "      <td>38780</td>\n",
       "      <td>bmw launch power model ever new competit packag boost power output m5 m6 new bmw m5 competit packag david motton bst may commentscom new bmw m5 m6 competit packag power road car bmw histori competit packag add output v8 fit mcar well power competit packag includ new suspens steer set new exhaust...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33907</th>\n",
       "      <td>62771</td>\n",
       "      <td>fantast four rise silver surfer pg part le fantastiqu seri appeal scienc fiction fan thrill seeker age film meet greatest challeng yet intergalact silver surfer race round globe wreak havoc mankind must unravel mysteri silver surfer confront mortal enemi dr doom hope lost shrek third u featur vo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3806</th>\n",
       "      <td>7073</td>\n",
       "      <td>russian ice star readi wow audienc grove theatr dunstabl snow white ice spectacular novemb success beauti beast ice cirqu de glace russian ice star return dazzl perform includ former world european olymp ice skate competitor theatric ice show stunningli choreograph ice spectacular transform stag...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12461</th>\n",
       "      <td>23025</td>\n",
       "      <td>sir alex ferguson nt normal guarante manchest unit reach champion leagu final may one name cert team sheet paul schole alreadi hand place final fergi red beat lyon tomorrow success reach final schole miss triumphant final barcelona bayern munich due suspens pick second yellow card semi juventu p...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7792</th>\n",
       "      <td>14411</td>\n",
       "      <td>zodiac judo club continu rich histori success foremost counti competit local outfit squad return sussex counti junior senior close trial worth six gold six silver four bronz medal four junior two senior player took first place due repres sussex southern area intercounti team championship sunday ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       index  \\\n",
       "2457    4536   \n",
       "37084  68626   \n",
       "14392  26626   \n",
       "21087  39028   \n",
       "33050  61161   \n",
       "39541  73204   \n",
       "18807  34808   \n",
       "25285  46753   \n",
       "4557    8477   \n",
       "30239  55940   \n",
       "15881  29377   \n",
       "14662  27150   \n",
       "31811  58837   \n",
       "12307  22722   \n",
       "7835   14483   \n",
       "20950  38780   \n",
       "33907  62771   \n",
       "3806    7073   \n",
       "12461  23025   \n",
       "7792   14411   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                              text  \n",
       "2457   everyon knew week would special one life sport bond guarante jackpot offer member draw eagerli anticip time appreci memor week incred jackpot winner found first draw number winner noel aiken member boyd dougla agenc genial noel member sinc week one sport bond nine year ago former gallah employe ...  \n",
       "37084  add stori itv sport involv major polit confront tv giant al jazeera could cost english network multimillion pound champion leagu product contract middl east al jazeera present gari linek arab station bbc gari linek present widespread coverag competit new sport channel itv sport programm product ...  \n",
       "14392  david suchet apo poirot final board orient express year play sleuth david suchet tell naomi west version excit disturb david suchet star agatha christi apo poirot murder orient express christma day david suchet star agatha christi apo poirot murder orient express christma day photo itv naomi wes...  \n",
       "21087  come plantat peac question pose new north west busi leader chamber agm last week jim sammon join rank host high profil busi champion led chamber commerc movement nw incom presid jim laid key prioriti two year term offic clear whilst derri region capit citi therefor chamber region leader new pres...  \n",
       "33050  teenag row die drag yard polic station stab wound daili mail report last updat pm septemb add stori victim craig marshal teenag die result violenc london year teenag die walk polic station blood pour stab wound craig marshal knife repeatedli row manag stagger yard acton polic station west london...  \n",
       "39541  monday decemb oxford utd oxford mail photograph author david pritchard chief sport report cover oxford unit follow us twitter oxfordmailoufc call oxford mail rodney parad play surfac morn pictur mark jone rodney parad play surfac morn pictur mark jone wet weather wale cast doubt oxford unit visi...  \n",
       "18807  polic offic accus collud colleagu arrest young protest fals ground student fee demonstr polic watchdog said investig seriou alleg offic abus posit breach cordon highprofil demonstr parliament squar last decemb rachel cerfontyn commission independ polic complaint commiss said also look circumst m...  \n",
       "25285  vladimir putin ever take anyth worri replac someth bigger better address alleg stole ring billionair robert kraft award american footbal team own super bowl russian presid said know rememb neither mr kraft ring valuabl thing kraft team propos putin said respons question annual econom forum st pe...  \n",
       "4557   jenni perciv westminst editor gordon brown scottish mp exclud vote englandonli law radic conserv plan unveil week tori leader david cameron pois back major overhaul westminst system english mp would meet separ grand committe decid legisl scotland mp would expect sit full vote common would speake...  \n",
       "30239  rsa insur rose tumultu insur sector activist investor appear compani sharehold regist share beleagu group tuesday start deepli discount right issu advanc cevian capit disclos stake compani swedish investor selfdescrib activist success push demerg engin cookson previous agit chang reinsur munich ...  \n",
       "15881  david cameron travel brussel eu summit help endors place munich articl time professor econom univers munich hanswern sinn prais prime minist recent speech eu essenc mr cameron right someth amiss eu regul far mani thing fall outsid remit prime minist new fan germani appear may struggl convinc adv...  \n",
       "14662  secondjob offic manag owe tax may liz keen western mail compani hous manag took second job make end meet given suspend jail sentenc admit evad incom tax john llewellyn end owe inland revenu seven year parttim deliveri driver wh smith cardiff crown court heard start retail earn inform technolog m...  \n",
       "31811  publish briton ever christma shop onlin decemb almost one five nonfood purchas made internet web sale growth acceler compar month fastest rate three year overal uk retail sale grew likeforlik basi data british retail consortium brc survey carri kpmg show onlin trade repres total nonfood sale dec...  \n",
       "12307  man charg connect threehour bank sieg polic said around peopl taken hostag monday barclay bank church road ashford west london matthew nutley brooksid avenu ashford charg robberi fals imprison possess imit firearm intent commit robberi two count make threat kill also charg blackmail possess offe...  \n",
       "7835   investig claim doctor grant women illeg abort base sex unborn babi launch depart health spark undercov newspap investig sexselect abort secretli film doctor british clinic agre termin foetus either male femal doctor also allegedli record admit prepar falsifi paperwork arrang illeg abort health s...  \n",
       "20950  bmw launch power model ever new competit packag boost power output m5 m6 new bmw m5 competit packag david motton bst may commentscom new bmw m5 m6 competit packag power road car bmw histori competit packag add output v8 fit mcar well power competit packag includ new suspens steer set new exhaust...  \n",
       "33907  fantast four rise silver surfer pg part le fantastiqu seri appeal scienc fiction fan thrill seeker age film meet greatest challeng yet intergalact silver surfer race round globe wreak havoc mankind must unravel mysteri silver surfer confront mortal enemi dr doom hope lost shrek third u featur vo...  \n",
       "3806   russian ice star readi wow audienc grove theatr dunstabl snow white ice spectacular novemb success beauti beast ice cirqu de glace russian ice star return dazzl perform includ former world european olymp ice skate competitor theatric ice show stunningli choreograph ice spectacular transform stag...  \n",
       "12461  sir alex ferguson nt normal guarante manchest unit reach champion leagu final may one name cert team sheet paul schole alreadi hand place final fergi red beat lyon tomorrow success reach final schole miss triumphant final barcelona bayern munich due suspens pick second yellow card semi juventu p...  \n",
       "7792   zodiac judo club continu rich histori success foremost counti competit local outfit squad return sussex counti junior senior close trial worth six gold six silver four bronz medal four junior two senior player took first place due repres sussex southern area intercounti team championship sunday ...  "
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Now examine hard and soft classifications manually\n",
    "import random\n",
    "\n",
    "pd.set_option('display.max_colwidth', 300)\n",
    "hardsdf = sdf.loc[y_test_pred==1, ['text']]\n",
    "hardsdf=hardsdf.reset_index()\n",
    "hardsdf.loc[random.sample(range(len(hardsdf['text'])), 20)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Soft News"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index</th>\n",
       "      <th>text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>6902</th>\n",
       "      <td>15012</td>\n",
       "      <td>london olymp new zealand apo black haka dare unleash morri dancer film director danni boyl eye new zealand apo rugbi world cup ahead plan olymp game open ceremoni forev england could morri men prove olymp hit london olymp new zealand apo black haka dare unleash morri dancer forev england could m...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22831</th>\n",
       "      <td>49708</td>\n",
       "      <td>ethan mcnern publish monday januari japan new prime minist visit tokyo main shinto shrine strong imperi connect appar attempt appeal rightw support shinzo abe recent start second stint japan leader visit meiji shrine yesterday becom first premier sinc visit shrine commemor emperor meiji symbol j...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27008</th>\n",
       "      <td>58795</td>\n",
       "      <td>saturday night wont without twohour fix scandinavian intrigu anyon would said id get hook norwegian polit drama complet subtitl wouldnt believ borgen fit seamlessli spot vacat kill might fashion piec knitwear sight ye comparison west wing borgen huge enjoy chart rise wonder brigitt nyborg unlik ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6530</th>\n",
       "      <td>14204</td>\n",
       "      <td>villag warn suspect util man publish date may locat eastbourn villag warn suspect util manhttp wwweastbourneheraldcoukviewarticleaspx articleid2858540 sectionid488</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23984</th>\n",
       "      <td>52281</td>\n",
       "      <td>watch clip bob godfrey famou cartoon roobarb continu read main stori relat stori actor richard brier die age anim muchlov cartoon roobarb henri cat die age famili confirm bob godfrey oscar short film great biographi engin isambard kingdom brunel remain best known tv work british anim whose caree...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17047</th>\n",
       "      <td>37052</td>\n",
       "      <td>game fa cup upset written truth walsal exit famou competit first hurdl manag chri hutch desper need win cool intens pressur surround posit got result left travel fan equal unhappi game alway go tricki encount leagu one basement boy despit sit place higher blue squar premier leagu oppon averag su...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10729</th>\n",
       "      <td>23398</td>\n",
       "      <td>marcu horan reveal hurt behind ireland defeat twickenham last year drive side croke park tomorrow night grand slam grab irish clash martin johnson men eagerli anticip least declan kidney player seek reveng ireland put england sword croke park two year ago last year went game signal end eddi osul...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32709</th>\n",
       "      <td>71208</td>\n",
       "      <td>lancast reliev fouri stay wednesday januari hendr fouri eight cap england hendr fouri eight cap england photo enlarg england head coach stuart lancast express relief hendr fouri stay unit kingdom fear could deport south africaborn flanker told leav countri despit cap eight time england contract ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28627</th>\n",
       "      <td>62260</td>\n",
       "      <td>jul bright children middl work class famili miss profession job continu elit governmentcommiss report warn report crossparti panel chair former cabinet minist alan milburn call urgent action break close shop mental say still characteris profess britain panel found half top profession job still t...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22555</th>\n",
       "      <td>49121</td>\n",
       "      <td>queen honour covet icon award bmi award london tuesday night bandmemb brian may roger taylor hand collect trophi event held dorchest hotel celebr musician whose song perform america last year honour come within week would late frontman freddi mercuri apo birthday last month group also celebr ann...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20666</th>\n",
       "      <td>45052</td>\n",
       "      <td>glaxosmithklin settl drug disput glaxosmithklin set pay one pharmaceut industri largest ever fine resolv disput us govern way market develop drug cost rachel cooper rachel cooper gmt nov commentscom britain biggest drug maker said thursday agre principl settl civil crimin claim final term expect...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2311</th>\n",
       "      <td>5019</td>\n",
       "      <td>secur firm updat facebook statu apo holiday burglar allegedli use facebook find target new servic kepp user apo social network profil updat apo holiday john timpson snoop facebook ever justifi recruit practic facebook allegedli use wouldb burglar photo pa matt warman matt warman consum technolog...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8499</th>\n",
       "      <td>18508</td>\n",
       "      <td>thursday april news darlington stockton time photograph author graem hetherington darlington stockton time bridgemast alan murray princip engin rodger wakerley sydney harbour bridg print bridgemast alan murray princip engin rodger wakerley sydney harbour bridg print darlington stockton time brid...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9659</th>\n",
       "      <td>21076</td>\n",
       "      <td>coupl flee hous set fire suspect arson attack pair asleep home barbel avenu eastrop basingstok neighbour spot blaze front hous morn fifteen firefight basingstok fire station attend scene extinguish blaze minut detect treat incid arson appeal wit detect constabl dan lees basingstok cid said serio...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32554</th>\n",
       "      <td>70863</td>\n",
       "      <td>celebr year hospit radio http iccheshireonlineicnetworkcouk0100news0100regionalnewstmheadlinecelebrating40yearsofhospitalradio</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4748</th>\n",
       "      <td>10320</td>\n",
       "      <td>soro sterl work davo forum friday januari share digg delici facebook stumbleupon twitter print email text size normal larg extra larg anoth day speech world econom forum davo switzerland brickbat uk economi specul georg soro warn prime minist david cameron recant harsh measur comprehens spend re...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24219</th>\n",
       "      <td>52808</td>\n",
       "      <td>never mind x factor widol women institut hope storm chart search talent member univers record home act ladi gaga ami winehous sign million deal recruit troup singer wi judg process select five hope nationwid search quietli launch earlier year quintet call harmoni due releas debut album octob wi ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23715</th>\n",
       "      <td>51645</td>\n",
       "      <td>dec janic burn doctor work scotland nh cage minimum yearstoday tri murderhundr holidaymak glasgow airport iraqi terrorist bilal abdulla plot indiscrimin wholesal murder airport busi london nightclub slouch dock show emot yesterday juri found guilti yesterday senior detect said took hippocrat oat...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33069</th>\n",
       "      <td>71979</td>\n",
       "      <td>http wwwjournallivecoukpafeedsnewcastlesportsnews20121128ainslietojoinsupportingcast8422932321758</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32064</th>\n",
       "      <td>69832</td>\n",
       "      <td>recent publish fastpac sherlock holm spoof hound baskervil come wokingham theatr tonight take arthur conan doyl stori featur three cast member group form promot wokingham mustvisit destin john redwood mp urg govern forc peac europ middl east previou articl famili call answer death aston mclean w...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       index  \\\n",
       "6902   15012   \n",
       "22831  49708   \n",
       "27008  58795   \n",
       "6530   14204   \n",
       "23984  52281   \n",
       "17047  37052   \n",
       "10729  23398   \n",
       "32709  71208   \n",
       "28627  62260   \n",
       "22555  49121   \n",
       "20666  45052   \n",
       "2311    5019   \n",
       "8499   18508   \n",
       "9659   21076   \n",
       "32554  70863   \n",
       "4748   10320   \n",
       "24219  52808   \n",
       "23715  51645   \n",
       "33069  71979   \n",
       "32064  69832   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                              text  \n",
       "6902   london olymp new zealand apo black haka dare unleash morri dancer film director danni boyl eye new zealand apo rugbi world cup ahead plan olymp game open ceremoni forev england could morri men prove olymp hit london olymp new zealand apo black haka dare unleash morri dancer forev england could m...  \n",
       "22831  ethan mcnern publish monday januari japan new prime minist visit tokyo main shinto shrine strong imperi connect appar attempt appeal rightw support shinzo abe recent start second stint japan leader visit meiji shrine yesterday becom first premier sinc visit shrine commemor emperor meiji symbol j...  \n",
       "27008  saturday night wont without twohour fix scandinavian intrigu anyon would said id get hook norwegian polit drama complet subtitl wouldnt believ borgen fit seamlessli spot vacat kill might fashion piec knitwear sight ye comparison west wing borgen huge enjoy chart rise wonder brigitt nyborg unlik ...  \n",
       "6530                                                                                                                                           villag warn suspect util man publish date may locat eastbourn villag warn suspect util manhttp wwweastbourneheraldcoukviewarticleaspx articleid2858540 sectionid488  \n",
       "23984  watch clip bob godfrey famou cartoon roobarb continu read main stori relat stori actor richard brier die age anim muchlov cartoon roobarb henri cat die age famili confirm bob godfrey oscar short film great biographi engin isambard kingdom brunel remain best known tv work british anim whose caree...  \n",
       "17047  game fa cup upset written truth walsal exit famou competit first hurdl manag chri hutch desper need win cool intens pressur surround posit got result left travel fan equal unhappi game alway go tricki encount leagu one basement boy despit sit place higher blue squar premier leagu oppon averag su...  \n",
       "10729  marcu horan reveal hurt behind ireland defeat twickenham last year drive side croke park tomorrow night grand slam grab irish clash martin johnson men eagerli anticip least declan kidney player seek reveng ireland put england sword croke park two year ago last year went game signal end eddi osul...  \n",
       "32709  lancast reliev fouri stay wednesday januari hendr fouri eight cap england hendr fouri eight cap england photo enlarg england head coach stuart lancast express relief hendr fouri stay unit kingdom fear could deport south africaborn flanker told leav countri despit cap eight time england contract ...  \n",
       "28627  jul bright children middl work class famili miss profession job continu elit governmentcommiss report warn report crossparti panel chair former cabinet minist alan milburn call urgent action break close shop mental say still characteris profess britain panel found half top profession job still t...  \n",
       "22555  queen honour covet icon award bmi award london tuesday night bandmemb brian may roger taylor hand collect trophi event held dorchest hotel celebr musician whose song perform america last year honour come within week would late frontman freddi mercuri apo birthday last month group also celebr ann...  \n",
       "20666  glaxosmithklin settl drug disput glaxosmithklin set pay one pharmaceut industri largest ever fine resolv disput us govern way market develop drug cost rachel cooper rachel cooper gmt nov commentscom britain biggest drug maker said thursday agre principl settl civil crimin claim final term expect...  \n",
       "2311   secur firm updat facebook statu apo holiday burglar allegedli use facebook find target new servic kepp user apo social network profil updat apo holiday john timpson snoop facebook ever justifi recruit practic facebook allegedli use wouldb burglar photo pa matt warman matt warman consum technolog...  \n",
       "8499   thursday april news darlington stockton time photograph author graem hetherington darlington stockton time bridgemast alan murray princip engin rodger wakerley sydney harbour bridg print bridgemast alan murray princip engin rodger wakerley sydney harbour bridg print darlington stockton time brid...  \n",
       "9659   coupl flee hous set fire suspect arson attack pair asleep home barbel avenu eastrop basingstok neighbour spot blaze front hous morn fifteen firefight basingstok fire station attend scene extinguish blaze minut detect treat incid arson appeal wit detect constabl dan lees basingstok cid said serio...  \n",
       "32554                                                                                                                                                                               celebr year hospit radio http iccheshireonlineicnetworkcouk0100news0100regionalnewstmheadlinecelebrating40yearsofhospitalradio  \n",
       "4748   soro sterl work davo forum friday januari share digg delici facebook stumbleupon twitter print email text size normal larg extra larg anoth day speech world econom forum davo switzerland brickbat uk economi specul georg soro warn prime minist david cameron recant harsh measur comprehens spend re...  \n",
       "24219  never mind x factor widol women institut hope storm chart search talent member univers record home act ladi gaga ami winehous sign million deal recruit troup singer wi judg process select five hope nationwid search quietli launch earlier year quintet call harmoni due releas debut album octob wi ...  \n",
       "23715  dec janic burn doctor work scotland nh cage minimum yearstoday tri murderhundr holidaymak glasgow airport iraqi terrorist bilal abdulla plot indiscrimin wholesal murder airport busi london nightclub slouch dock show emot yesterday juri found guilti yesterday senior detect said took hippocrat oat...  \n",
       "33069                                                                                                                                                                                                            http wwwjournallivecoukpafeedsnewcastlesportsnews20121128ainslietojoinsupportingcast8422932321758  \n",
       "32064  recent publish fastpac sherlock holm spoof hound baskervil come wokingham theatr tonight take arthur conan doyl stori featur three cast member group form promot wokingham mustvisit destin john redwood mp urg govern forc peac europ middl east previou articl famili call answer death aston mclean w...  "
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "softsdf = sdf.loc[y_test_pred==0, ['text']]\n",
    "softsdf=softsdf.reset_index()\n",
    "softsdf.loc[random.sample(range(len(softsdf['text'])), 20)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save the vectorizer and classifier\n",
    "import pickle\n",
    "with open('./models/url_uk_classifier.pkl', 'wb') as fid:\n",
    "    pickle.dump(clf, fid)\n",
    "\n",
    "with open('./models/url_uk_vectorizer.pkl', 'wb') as fid:\n",
    "    pickle.dump(vectorizer, fid)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "with open('./models/url_uk_src_name_dummy.pkl', 'wb') as fid:\n",
    "    pickle.dump(ohenc, fid)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plotting Success"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix:\n",
      "Predicted    0.0    1.0  __all__\n",
      "Actual                          \n",
      "0.0        30461   5451    35912\n",
      "1.0         3791  34730    38521\n",
      "__all__    34252  40181    74433\n",
      "       0.0    1.0\n",
      "0.0  30461   5451\n",
      "1.0   3791  34730\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x576 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib\n",
    "matplotlib.style.use('ggplot')\n",
    "from pandas_confusion import ConfusionMatrix\n",
    "\n",
    "def get_most_informative_features(vectorizer, clf, n=20):\n",
    "    feature_names = vectorizer.get_feature_names()\n",
    "    coefs_with_fns = sorted(zip(clf.coef_[0], feature_names))\n",
    "    top_a = coefs_with_fns[:n]\n",
    "    top_b = coefs_with_fns[:-(n + 1):-1]\n",
    "    return top_a, top_b\n",
    "\n",
    "vect = vectorizer\n",
    "y_pred = y_test_pred\n",
    "\n",
    "top_a, top_b = get_most_informative_features(vect, clf, 100)\n",
    "top_a_df = pd.DataFrame(top_a)\n",
    "top_a_df.columns = ['coef', 'term']\n",
    "top_b_df = pd.DataFrame(top_b)\n",
    "top_b_df.columns = ['coef', 'term']\n",
    "\n",
    "top_a_df.to_csv('./reports/url_uk_top100_soft.csv', index=False)\n",
    "top_b_df.to_csv('./reports/url_uk_top100_hard.csv', index=False)\n",
    "\n",
    "y_test_df = pd.DataFrame(test_df['hard_soft'], columns=['actual']).reset_index(drop=True)\n",
    "y_test_df['predicted'] = y_pred\n",
    "y_test_df.to_csv('./tests/url_uk_test_prediction.csv', index=False)\n",
    "\n",
    "y_true = test_df['hard_soft'].values\n",
    "confusion_matrix = ConfusionMatrix(y_true, y_pred)\n",
    "print(\"Confusion matrix:\\n%s\" % confusion_matrix)\n",
    "\n",
    "confusion_matrix.plot()\n",
    "\n",
    "from sklearn.metrics import confusion_matrix\n",
    "conf = confusion_matrix(y_true, y_pred)\n",
    "conf_df = pd.DataFrame(conf)\n",
    "conf_df.columns = clf.classes_\n",
    "conf_df.index = clf.classes_\n",
    "conf_df.to_csv('./reports/url_uk_test_confusion_matrix.csv', index_label=\"actual \\ predicted\")\n",
    "print(conf_df)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Proportion of Hard and Soft"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f9b4d67a2b0>"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sdf.groupby(['hard_soft']).agg({'url': 'count'}).plot(kind='bar')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "37582\n"
     ]
    }
   ],
   "source": [
    "clf_pkl = 'models/url_uk_classifier.pkl'\n",
    "vect_pkl = 'models/url_uk_vectorizer.pkl'\n",
    "\n",
    "with open(clf_pkl, 'rb') as fid:\n",
    "    clf = pickle.load(fid)\n",
    "\n",
    "with open(vect_pkl, 'rb') as fid:\n",
    "    vect = pickle.load(fid)\n",
    "print(len(vect.vocabulary_))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n",
      "1\n",
      "2\n",
      "3\n",
      "4\n",
      "5\n",
      "6\n",
      "7\n",
      "8\n",
      "9\n",
      "10\n",
      "11\n",
      "12\n",
      "13\n",
      "14\n",
      "15\n",
      "16\n",
      "17\n",
      "18\n",
      "19\n",
      "20\n",
      "21\n",
      "22\n",
      "23\n",
      "24\n",
      "25\n",
      "26\n",
      "27\n",
      "28\n",
      "29\n",
      "30\n",
      "31\n",
      "32\n",
      "33\n",
      "34\n",
      "35\n",
      "36\n",
      "37\n",
      "38\n",
      "39\n",
      "40\n",
      "41\n",
      "42\n",
      "43\n",
      "44\n",
      "45\n",
      "46\n",
      "47\n",
      "48\n",
      "49\n",
      "50\n",
      "51\n",
      "52\n",
      "53\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_extraction.text import TfidfTransformer\n",
    "\n",
    "out_df = pd.DataFrame()\n",
    "i = 0\n",
    "for df in pd.read_csv('/opt/notebooks/not_news/tabs/uk_news_media_data_subset_by_label.csv.gz', usecols=['date', 'sourceName', 'text', 'url', 'domain', 'label', 'articleId'], chunksize=100000, iterator=True):\n",
    "    df['src_name'] = df['sourceName'].str.lower()\n",
    "    del df['sourceName']\n",
    "    X = vect.transform(df.text.astype(str))\n",
    "    tfidf = TfidfTransformer()\n",
    "    X = tfidf.fit_transform(X)\n",
    "    y_pred = clf.predict(X)\n",
    "    y_prob = clf.predict_proba(X)\n",
    "    df['pred_label'] = y_pred\n",
    "    df.loc[df.url.notnull(), 'soft_lab'] = df.loc[df.url.notnull(), 'url'].apply(lambda c: is_soft_lab(c))\n",
    "    df.loc[df.url.notnull(), 'hard_lab'] = df.loc[df.url.notnull(), 'url'].apply(lambda c: is_hard_lab(c))\n",
    "    df['pred_prob'] = y_prob[:, 1]\n",
    "    df.loc[(df.hard_lab == 1) & df.soft_lab.isnull(), 'train_label'] = 1\n",
    "    df.loc[df.hard_lab.isnull() & (df.soft_lab == 1), 'train_label'] = 0\n",
    "    del df['soft_lab']\n",
    "    del df['hard_lab']\n",
    "    del df['text']\n",
    "    out_df = out_df.append(df)\n",
    "    del df\n",
    "    del tfidf\n",
    "    del X\n",
    "    #print(i, out_df.memory_usage(deep=True))\n",
    "    print(i)\n",
    "    gc.collect()\n",
    "    i += 1\n",
    "    #if i > 10:\n",
    "    #    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>url</th>\n",
       "      <th>articleId</th>\n",
       "      <th>date</th>\n",
       "      <th>domain</th>\n",
       "      <th>label</th>\n",
       "      <th>src_name</th>\n",
       "      <th>pred_label</th>\n",
       "      <th>pred_prob</th>\n",
       "      <th>train_label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&amp;category=NEWS&amp;tBrand=bbjonline&amp;tCategory=news&amp;itemid=NOED23%20Dec%202008%2009%3A41%3A39%3A63</td>\n",
       "      <td>en_20081223_becclesan6437_379770410001</td>\n",
       "      <td>2008-12-22</td>\n",
       "      <td>becclesandbungayjournal.co.uk</td>\n",
       "      <td>Beccles and Bungay Journal</td>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>0</td>\n",
       "      <td>0.447541</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&amp;category=NEWS&amp;tBrand=bbjonline&amp;tCategory=news&amp;itemid=NOED08%20Jan%202010%2012%3A37%3A14%3A71</td>\n",
       "      <td>en_20100108_becclesan6437_844822170000</td>\n",
       "      <td>2010-01-08</td>\n",
       "      <td>becclesandbungayjournal.co.uk</td>\n",
       "      <td>Beccles and Bungay Journal</td>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>1</td>\n",
       "      <td>0.583057</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&amp;category=NEWS&amp;tBrand=bbjonline&amp;tCategory=news&amp;itemid=NOED09%20Jun%202008%2010%3A09%3A10%3A98</td>\n",
       "      <td>en_20080609_becclesan6437_177751570001</td>\n",
       "      <td>2008-06-09</td>\n",
       "      <td>becclesandbungayjournal.co.uk</td>\n",
       "      <td>Beccles and Bungay Journal</td>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>1</td>\n",
       "      <td>0.790210</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&amp;category=NEWS&amp;tBrand=bbjonline&amp;tCategory=news&amp;itemid=NOED03%20Nov%202008%2010%3A03%3A33%3A76</td>\n",
       "      <td>en_20081103_becclesan6437_328210310001</td>\n",
       "      <td>2008-11-03</td>\n",
       "      <td>becclesandbungayjournal.co.uk</td>\n",
       "      <td>Beccles and Bungay Journal</td>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>1</td>\n",
       "      <td>0.809476</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>http://www.becclesandbungayjournal.co.uk/content/bbjournal/sport/story.aspx?brand=BBJOnline&amp;category=SPORT&amp;tBrand=bbjonline&amp;tCategory=sport&amp;itemid=NOED15%20May%202009%2010%3A52%3A14%3A91</td>\n",
       "      <td>en_20090515_becclesan6437_534951120000</td>\n",
       "      <td>2009-05-15</td>\n",
       "      <td>becclesandbungayjournal.co.uk</td>\n",
       "      <td>Beccles and Bungay Journal</td>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>0</td>\n",
       "      <td>0.193729</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                                                                                          url  \\\n",
       "0     http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&category=NEWS&tBrand=bbjonline&tCategory=news&itemid=NOED23%20Dec%202008%2009%3A41%3A39%3A63   \n",
       "1     http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&category=NEWS&tBrand=bbjonline&tCategory=news&itemid=NOED08%20Jan%202010%2012%3A37%3A14%3A71   \n",
       "2     http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&category=NEWS&tBrand=bbjonline&tCategory=news&itemid=NOED09%20Jun%202008%2010%3A09%3A10%3A98   \n",
       "3     http://www.becclesandbungayjournal.co.uk/content/bbjournal/news/story.aspx?brand=BBJOnline&category=NEWS&tBrand=bbjonline&tCategory=news&itemid=NOED03%20Nov%202008%2010%3A03%3A33%3A76   \n",
       "4  http://www.becclesandbungayjournal.co.uk/content/bbjournal/sport/story.aspx?brand=BBJOnline&category=SPORT&tBrand=bbjonline&tCategory=sport&itemid=NOED15%20May%202009%2010%3A52%3A14%3A91   \n",
       "\n",
       "                                articleId        date  \\\n",
       "0  en_20081223_becclesan6437_379770410001  2008-12-22   \n",
       "1  en_20100108_becclesan6437_844822170000  2010-01-08   \n",
       "2  en_20080609_becclesan6437_177751570001  2008-06-09   \n",
       "3  en_20081103_becclesan6437_328210310001  2008-11-03   \n",
       "4  en_20090515_becclesan6437_534951120000  2009-05-15   \n",
       "\n",
       "                          domain                       label  \\\n",
       "0  becclesandbungayjournal.co.uk  Beccles and Bungay Journal   \n",
       "1  becclesandbungayjournal.co.uk  Beccles and Bungay Journal   \n",
       "2  becclesandbungayjournal.co.uk  Beccles and Bungay Journal   \n",
       "3  becclesandbungayjournal.co.uk  Beccles and Bungay Journal   \n",
       "4  becclesandbungayjournal.co.uk  Beccles and Bungay Journal   \n",
       "\n",
       "                     src_name  pred_label  pred_prob  train_label  \n",
       "0  beccles and bungay journal           0   0.447541          NaN  \n",
       "1  beccles and bungay journal           1   0.583057          NaN  \n",
       "2  beccles and bungay journal           1   0.790210          NaN  \n",
       "3  beccles and bungay journal           1   0.809476          NaN  \n",
       "4  beccles and bungay journal           0   0.193729          0.0  "
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "out_df.loc[:, 'pred_label'] = out_df.loc[:, 'pred_label'].astype(int)\n",
    "out_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "out_df.to_csv('./predicts/uk_media_url_pred.csv', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>url</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>src_name</th>\n",
       "      <th>pred_label</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">24dash</th>\n",
       "      <th>0</th>\n",
       "      <td>2983</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>40669</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">a world to win blogs</th>\n",
       "      <th>0</th>\n",
       "      <td>180</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1369</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">aberdeen independent</th>\n",
       "      <th>0</th>\n",
       "      <td>145</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">agra-net.com</th>\n",
       "      <th>0</th>\n",
       "      <td>6516</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7875</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">al-sharq al-awsat</th>\n",
       "      <th>0</th>\n",
       "      <td>1044</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>15194</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">alert net</th>\n",
       "      <th>0</th>\n",
       "      <td>597</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>14548</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">alliance for workers'' liberty</th>\n",
       "      <th>0</th>\n",
       "      <td>304</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1477</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ananova</th>\n",
       "      <th>0</th>\n",
       "      <td>12617</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>66280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ananova - orange news</th>\n",
       "      <th>0</th>\n",
       "      <td>3823</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4173</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">andersontown news</th>\n",
       "      <th>0</th>\n",
       "      <td>9859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>30673</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">andover advertiser</th>\n",
       "      <th>0</th>\n",
       "      <td>2115</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">asharq al-awsat</th>\n",
       "      <th>0</th>\n",
       "      <td>79</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2681</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">asharq alawsatt</th>\n",
       "      <th>0</th>\n",
       "      <td>490</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>16184</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ayrshire post</th>\n",
       "      <th>0</th>\n",
       "      <td>2874</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6328</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">bakery and snacks</th>\n",
       "      <th>0</th>\n",
       "      <td>2430</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1361</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">west sussex county times</th>\n",
       "      <th>0</th>\n",
       "      <td>1911</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1513</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">western mail</th>\n",
       "      <th>0</th>\n",
       "      <td>25531</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>50173</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wharf</th>\n",
       "      <th>0</th>\n",
       "      <td>2935</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2191</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wiltshire news / this is wiltshire</th>\n",
       "      <th>0</th>\n",
       "      <td>5293</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>11244</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wimbledon guardian</th>\n",
       "      <th>0</th>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>169</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wirral globe</th>\n",
       "      <th>0</th>\n",
       "      <td>605</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1861</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wisbech standard</th>\n",
       "      <th>0</th>\n",
       "      <td>1719</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3387</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">woking news and mail</th>\n",
       "      <th>0</th>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>521</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wokingham times</th>\n",
       "      <th>0</th>\n",
       "      <td>5289</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5638</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">worcester news</th>\n",
       "      <th>0</th>\n",
       "      <td>4081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5879</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">worksop guardian</th>\n",
       "      <th>0</th>\n",
       "      <td>2514</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">world nuclear news</th>\n",
       "      <th>0</th>\n",
       "      <td>31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">worthing herald</th>\n",
       "      <th>0</th>\n",
       "      <td>1003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2093</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">yorkshire evening post</th>\n",
       "      <th>0</th>\n",
       "      <td>5895</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>13920</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">yorkshire post</th>\n",
       "      <th>0</th>\n",
       "      <td>232</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>780</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>688 rows × 1 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                 url\n",
       "src_name                           pred_label       \n",
       "24dash                             0            2983\n",
       "                                   1           40669\n",
       "a world to win blogs               0             180\n",
       "                                   1            1369\n",
       "aberdeen independent               0             145\n",
       "                                   1              42\n",
       "agra-net.com                       0            6516\n",
       "                                   1            7875\n",
       "al-sharq al-awsat                  0            1044\n",
       "                                   1           15194\n",
       "alert net                          0             597\n",
       "                                   1           14548\n",
       "alliance for workers'' liberty     0             304\n",
       "                                   1            1477\n",
       "ananova                            0           12617\n",
       "                                   1           66280\n",
       "ananova - orange news              0            3823\n",
       "                                   1            4173\n",
       "andersontown news                  0            9859\n",
       "                                   1           30673\n",
       "andover advertiser                 0            2115\n",
       "                                   1            5425\n",
       "asharq al-awsat                    0              79\n",
       "                                   1            2681\n",
       "asharq alawsatt                    0             490\n",
       "                                   1           16184\n",
       "ayrshire post                      0            2874\n",
       "                                   1            6328\n",
       "bakery and snacks                  0            2430\n",
       "                                   1            1361\n",
       "...                                              ...\n",
       "west sussex county times           0            1911\n",
       "                                   1            1513\n",
       "western mail                       0           25531\n",
       "                                   1           50173\n",
       "wharf                              0            2935\n",
       "                                   1            2191\n",
       "wiltshire news / this is wiltshire 0            5293\n",
       "                                   1           11244\n",
       "wimbledon guardian                 0              41\n",
       "                                   1             169\n",
       "wirral globe                       0             605\n",
       "                                   1            1861\n",
       "wisbech standard                   0            1719\n",
       "                                   1            3387\n",
       "woking news and mail               0              45\n",
       "                                   1             521\n",
       "wokingham times                    0            5289\n",
       "                                   1            5638\n",
       "worcester news                     0            4081\n",
       "                                   1            5879\n",
       "worksop guardian                   0            2514\n",
       "                                   1            1336\n",
       "world nuclear news                 0              31\n",
       "                                   1            1010\n",
       "worthing herald                    0            1003\n",
       "                                   1            2093\n",
       "yorkshire evening post             0            5895\n",
       "                                   1           13920\n",
       "yorkshire post                     0             232\n",
       "                                   1             780\n",
       "\n",
       "[688 rows x 1 columns]"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "out_df.groupby(['src_name', 'pred_label']).agg({'url': 'count'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>url</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>label</th>\n",
       "      <th>pred_label</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">24dash</th>\n",
       "      <th>0</th>\n",
       "      <td>2983</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>40669</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">A World to Win blogs</th>\n",
       "      <th>0</th>\n",
       "      <td>180</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1369</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Agra-net.com</th>\n",
       "      <th>0</th>\n",
       "      <td>6516</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7875</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Al-Sharq al-Awsat</th>\n",
       "      <th>0</th>\n",
       "      <td>1613</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>34059</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Alliance for Workers' Liberty</th>\n",
       "      <th>0</th>\n",
       "      <td>304</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1477</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Ananova</th>\n",
       "      <th>0</th>\n",
       "      <td>16440</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>70453</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Andover Advertiser</th>\n",
       "      <th>0</th>\n",
       "      <td>2115</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Ayrshire Post</th>\n",
       "      <th>0</th>\n",
       "      <td>2874</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6328</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">BBC News</th>\n",
       "      <th>0</th>\n",
       "      <td>19852</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>100920</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Bakery and Snacks</th>\n",
       "      <th>0</th>\n",
       "      <td>2442</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Ballyclare Gazette</th>\n",
       "      <th>0</th>\n",
       "      <td>676</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>570</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Ballymena Times</th>\n",
       "      <th>0</th>\n",
       "      <td>2603</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1735</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Ballymoney Today</th>\n",
       "      <th>0</th>\n",
       "      <td>13945</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>11688</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Banbridge Leader</th>\n",
       "      <th>0</th>\n",
       "      <td>18395</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>16416</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Banbury Guardian</th>\n",
       "      <th>0</th>\n",
       "      <td>9535</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>27755</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">West Sussex County Times</th>\n",
       "      <th>0</th>\n",
       "      <td>2139</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1939</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Western Mail</th>\n",
       "      <th>0</th>\n",
       "      <td>25531</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>50173</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Wharf</th>\n",
       "      <th>0</th>\n",
       "      <td>2935</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2191</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Wiltshire News / This Is Wiltshire</th>\n",
       "      <th>0</th>\n",
       "      <td>5293</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>11244</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Wirral Globe</th>\n",
       "      <th>0</th>\n",
       "      <td>605</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1861</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Wisbech Standard</th>\n",
       "      <th>0</th>\n",
       "      <td>1719</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3387</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Wokingham Times</th>\n",
       "      <th>0</th>\n",
       "      <td>5289</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5638</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Worcester News</th>\n",
       "      <th>0</th>\n",
       "      <td>4081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5879</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Worksop Guardian</th>\n",
       "      <th>0</th>\n",
       "      <td>2514</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">World Nuclear News</th>\n",
       "      <th>0</th>\n",
       "      <td>31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Worthing Herald</th>\n",
       "      <th>0</th>\n",
       "      <td>1003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2093</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Yorkshire Evening Post</th>\n",
       "      <th>0</th>\n",
       "      <td>5895</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>13920</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Yorkshire Post</th>\n",
       "      <th>0</th>\n",
       "      <td>232</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>780</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Your Local Guardian</th>\n",
       "      <th>0</th>\n",
       "      <td>1652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7505</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">enfieldindependent.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>3683</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1967</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>552 rows × 1 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                  url\n",
       "label                              pred_label        \n",
       "24dash                             0             2983\n",
       "                                   1            40669\n",
       "A World to Win blogs               0              180\n",
       "                                   1             1369\n",
       "Agra-net.com                       0             6516\n",
       "                                   1             7875\n",
       "Al-Sharq al-Awsat                  0             1613\n",
       "                                   1            34059\n",
       "Alliance for Workers' Liberty      0              304\n",
       "                                   1             1477\n",
       "Ananova                            0            16440\n",
       "                                   1            70453\n",
       "Andover Advertiser                 0             2115\n",
       "                                   1             5425\n",
       "Ayrshire Post                      0             2874\n",
       "                                   1             6328\n",
       "BBC News                           0            19852\n",
       "                                   1           100920\n",
       "Bakery and Snacks                  0             2442\n",
       "                                   1             1365\n",
       "Ballyclare Gazette                 0              676\n",
       "                                   1              570\n",
       "Ballymena Times                    0             2603\n",
       "                                   1             1735\n",
       "Ballymoney Today                   0            13945\n",
       "                                   1            11688\n",
       "Banbridge Leader                   0            18395\n",
       "                                   1            16416\n",
       "Banbury Guardian                   0             9535\n",
       "                                   1            27755\n",
       "...                                               ...\n",
       "West Sussex County Times           0             2139\n",
       "                                   1             1939\n",
       "Western Mail                       0            25531\n",
       "                                   1            50173\n",
       "Wharf                              0             2935\n",
       "                                   1             2191\n",
       "Wiltshire News / This Is Wiltshire 0             5293\n",
       "                                   1            11244\n",
       "Wirral Globe                       0              605\n",
       "                                   1             1861\n",
       "Wisbech Standard                   0             1719\n",
       "                                   1             3387\n",
       "Wokingham Times                    0             5289\n",
       "                                   1             5638\n",
       "Worcester News                     0             4081\n",
       "                                   1             5879\n",
       "Worksop Guardian                   0             2514\n",
       "                                   1             1336\n",
       "World Nuclear News                 0               31\n",
       "                                   1             1010\n",
       "Worthing Herald                    0             1003\n",
       "                                   1             2093\n",
       "Yorkshire Evening Post             0             5895\n",
       "                                   1            13920\n",
       "Yorkshire Post                     0              232\n",
       "                                   1              780\n",
       "Your Local Guardian                0             1652\n",
       "                                   1             7505\n",
       "enfieldindependent.co.uk           0             3683\n",
       "                                   1             1967\n",
       "\n",
       "[552 rows x 1 columns]"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "out_df.groupby(['label', 'pred_label']).agg({'url': 'count'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>url</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain</th>\n",
       "      <th>pred_label</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">24dash.com</th>\n",
       "      <th>0</th>\n",
       "      <td>2983</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>40669</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">aawsat.com</th>\n",
       "      <th>0</th>\n",
       "      <td>450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2833</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">agra-net.com</th>\n",
       "      <th>0</th>\n",
       "      <td>6516</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7875</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">alertnet.org</th>\n",
       "      <th>0</th>\n",
       "      <td>1288</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>66438</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ananova.com</th>\n",
       "      <th>0</th>\n",
       "      <td>2039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>52624</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">andoveradvertiser.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>2115</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">asharq-e.com</th>\n",
       "      <th>0</th>\n",
       "      <td>796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>20280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">asharqalawsat.com</th>\n",
       "      <th>0</th>\n",
       "      <td>367</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10946</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">aworldtowin.net</th>\n",
       "      <th>0</th>\n",
       "      <td>180</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1369</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ayrshirepost.net</th>\n",
       "      <th>0</th>\n",
       "      <td>2729</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5999</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">bakeryandsnacks.com</th>\n",
       "      <th>0</th>\n",
       "      <td>2442</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ballyclaregazette.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>676</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>570</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ballymenatimes.com</th>\n",
       "      <th>0</th>\n",
       "      <td>2603</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1735</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ballymoneytimes.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>13945</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>11688</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">banbridgeleader.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>18395</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>16416</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">tyronetimes.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2937</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">walesonline.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>9070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>39774</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wandsworthguardian.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>5312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wharf.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>2189</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1934</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wirralglobe.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>605</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1861</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wisbechstandard.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>1719</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3387</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">worcesternews.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>4081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5879</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">workersliberty.org</th>\n",
       "      <th>0</th>\n",
       "      <td>304</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1477</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">worksopguardian.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>766</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>421</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">world-nuclear-news.org</th>\n",
       "      <th>0</th>\n",
       "      <td>31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">worthingherald.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>920</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wscountytimes.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>773</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>732</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">yorkpress.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>3916</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>8607</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">yorkshirepost.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>5895</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>13920</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">yourlocalguardian.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>1652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7505</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>582 rows × 1 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                       url\n",
       "domain                   pred_label       \n",
       "24dash.com               0            2983\n",
       "                         1           40669\n",
       "aawsat.com               0             450\n",
       "                         1            2833\n",
       "agra-net.com             0            6516\n",
       "                         1            7875\n",
       "alertnet.org             0            1288\n",
       "                         1           66438\n",
       "ananova.com              0            2039\n",
       "                         1           52624\n",
       "andoveradvertiser.co.uk  0            2115\n",
       "                         1            5425\n",
       "asharq-e.com             0             796\n",
       "                         1           20280\n",
       "asharqalawsat.com        0             367\n",
       "                         1           10946\n",
       "aworldtowin.net          0             180\n",
       "                         1            1369\n",
       "ayrshirepost.net         0            2729\n",
       "                         1            5999\n",
       "bakeryandsnacks.com      0            2442\n",
       "                         1            1365\n",
       "ballyclaregazette.co.uk  0             676\n",
       "                         1             570\n",
       "ballymenatimes.com       0            2603\n",
       "                         1            1735\n",
       "ballymoneytimes.co.uk    0           13945\n",
       "                         1           11688\n",
       "banbridgeleader.co.uk    0           18395\n",
       "                         1           16416\n",
       "...                                    ...\n",
       "tyronetimes.co.uk        0             743\n",
       "                         1            2937\n",
       "walesonline.co.uk        0            9070\n",
       "                         1           39774\n",
       "wandsworthguardian.co.uk 0            5312\n",
       "                         1            3336\n",
       "wharf.co.uk              0            2189\n",
       "                         1            1934\n",
       "wirralglobe.co.uk        0             605\n",
       "                         1            1861\n",
       "wisbechstandard.co.uk    0            1719\n",
       "                         1            3387\n",
       "worcesternews.co.uk      0            4081\n",
       "                         1            5879\n",
       "workersliberty.org       0             304\n",
       "                         1            1477\n",
       "worksopguardian.co.uk    0             766\n",
       "                         1             421\n",
       "world-nuclear-news.org   0              31\n",
       "                         1            1010\n",
       "worthingherald.co.uk     0             920\n",
       "                         1            1996\n",
       "wscountytimes.co.uk      0             773\n",
       "                         1             732\n",
       "yorkpress.co.uk          0            3916\n",
       "                         1            8607\n",
       "yorkshirepost.co.uk      0            5895\n",
       "                         1           13920\n",
       "yourlocalguardian.co.uk  0            1652\n",
       "                         1            7505\n",
       "\n",
       "[582 rows x 1 columns]"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "out_df.groupby(['domain', 'pred_label']).agg({'url': 'count'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>url</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain</th>\n",
       "      <th>pred_label</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">24dash.com</th>\n",
       "      <th>0</th>\n",
       "      <td>2983</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>40669</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">aawsat.com</th>\n",
       "      <th>0</th>\n",
       "      <td>450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2833</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">agra-net.com</th>\n",
       "      <th>0</th>\n",
       "      <td>6516</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7875</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">alertnet.org</th>\n",
       "      <th>0</th>\n",
       "      <td>1288</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>66438</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ananova.com</th>\n",
       "      <th>0</th>\n",
       "      <td>2039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>52624</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">andoveradvertiser.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>2115</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">asharq-e.com</th>\n",
       "      <th>0</th>\n",
       "      <td>796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>20280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">asharqalawsat.com</th>\n",
       "      <th>0</th>\n",
       "      <td>367</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10946</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">aworldtowin.net</th>\n",
       "      <th>0</th>\n",
       "      <td>180</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1369</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ayrshirepost.net</th>\n",
       "      <th>0</th>\n",
       "      <td>2729</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5999</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">bakeryandsnacks.com</th>\n",
       "      <th>0</th>\n",
       "      <td>2442</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ballyclaregazette.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>676</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>570</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ballymenatimes.com</th>\n",
       "      <th>0</th>\n",
       "      <td>2603</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1735</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">ballymoneytimes.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>13945</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>11688</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">banbridgeleader.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>18395</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>16416</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">tyronetimes.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2937</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">walesonline.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>9070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>39774</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wandsworthguardian.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>5312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wharf.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>2189</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1934</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wirralglobe.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>605</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1861</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wisbechstandard.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>1719</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3387</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">worcesternews.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>4081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5879</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">workersliberty.org</th>\n",
       "      <th>0</th>\n",
       "      <td>304</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1477</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">worksopguardian.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>766</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>421</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">world-nuclear-news.org</th>\n",
       "      <th>0</th>\n",
       "      <td>31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">worthingherald.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>920</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">wscountytimes.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>773</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>732</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">yorkpress.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>3916</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>8607</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">yorkshirepost.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>5895</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>13920</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">yourlocalguardian.co.uk</th>\n",
       "      <th>0</th>\n",
       "      <td>1652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7505</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>582 rows × 1 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                       url\n",
       "domain                   pred_label       \n",
       "24dash.com               0            2983\n",
       "                         1           40669\n",
       "aawsat.com               0             450\n",
       "                         1            2833\n",
       "agra-net.com             0            6516\n",
       "                         1            7875\n",
       "alertnet.org             0            1288\n",
       "                         1           66438\n",
       "ananova.com              0            2039\n",
       "                         1           52624\n",
       "andoveradvertiser.co.uk  0            2115\n",
       "                         1            5425\n",
       "asharq-e.com             0             796\n",
       "                         1           20280\n",
       "asharqalawsat.com        0             367\n",
       "                         1           10946\n",
       "aworldtowin.net          0             180\n",
       "                         1            1369\n",
       "ayrshirepost.net         0            2729\n",
       "                         1            5999\n",
       "bakeryandsnacks.com      0            2442\n",
       "                         1            1365\n",
       "ballyclaregazette.co.uk  0             676\n",
       "                         1             570\n",
       "ballymenatimes.com       0            2603\n",
       "                         1            1735\n",
       "ballymoneytimes.co.uk    0           13945\n",
       "                         1           11688\n",
       "banbridgeleader.co.uk    0           18395\n",
       "                         1           16416\n",
       "...                                    ...\n",
       "tyronetimes.co.uk        0             743\n",
       "                         1            2937\n",
       "walesonline.co.uk        0            9070\n",
       "                         1           39774\n",
       "wandsworthguardian.co.uk 0            5312\n",
       "                         1            3336\n",
       "wharf.co.uk              0            2189\n",
       "                         1            1934\n",
       "wirralglobe.co.uk        0             605\n",
       "                         1            1861\n",
       "wisbechstandard.co.uk    0            1719\n",
       "                         1            3387\n",
       "worcesternews.co.uk      0            4081\n",
       "                         1            5879\n",
       "workersliberty.org       0             304\n",
       "                         1            1477\n",
       "worksopguardian.co.uk    0             766\n",
       "                         1             421\n",
       "world-nuclear-news.org   0              31\n",
       "                         1            1010\n",
       "worthingherald.co.uk     0             920\n",
       "                         1            1996\n",
       "wscountytimes.co.uk      0             773\n",
       "                         1             732\n",
       "yorkpress.co.uk          0            3916\n",
       "                         1            8607\n",
       "yorkshirepost.co.uk      0            5895\n",
       "                         1           13920\n",
       "yourlocalguardian.co.uk  0            1652\n",
       "                         1            7505\n",
       "\n",
       "[582 rows x 1 columns]"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "out_df.groupby(['domain', 'pred_label']).agg({'url': 'count'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5392953, 9)"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "out_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn import linear_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "out_df['norm_date'] = pd.to_datetime(out_df['date'], format='%Y-%m-%d').astype(int) / 1000.0\n",
    "scaler = MinMaxScaler(feature_range=(0, 1))\n",
    "out_df['norm_date'] = scaler.fit_transform(out_df[['norm_date']])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>src_name</th>\n",
       "      <th>date</th>\n",
       "      <th>norm_date</th>\n",
       "      <th>pred_prob</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2008-12-22</td>\n",
       "      <td>0.493774</td>\n",
       "      <td>0.447541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2010-01-08</td>\n",
       "      <td>0.580258</td>\n",
       "      <td>0.583057</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2008-06-09</td>\n",
       "      <td>0.449400</td>\n",
       "      <td>0.790210</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2008-11-03</td>\n",
       "      <td>0.482681</td>\n",
       "      <td>0.809476</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2009-05-15</td>\n",
       "      <td>0.526375</td>\n",
       "      <td>0.193729</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2007-05-24</td>\n",
       "      <td>0.362916</td>\n",
       "      <td>0.512569</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2009-03-28</td>\n",
       "      <td>0.515508</td>\n",
       "      <td>0.020062</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2010-01-13</td>\n",
       "      <td>0.581390</td>\n",
       "      <td>0.207929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2007-07-26</td>\n",
       "      <td>0.377179</td>\n",
       "      <td>0.597845</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2009-12-16</td>\n",
       "      <td>0.575051</td>\n",
       "      <td>0.382439</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2010-07-05</td>\n",
       "      <td>0.620557</td>\n",
       "      <td>0.232942</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2007-11-23</td>\n",
       "      <td>0.404347</td>\n",
       "      <td>0.057177</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2007-12-08</td>\n",
       "      <td>0.407743</td>\n",
       "      <td>0.683320</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2009-09-11</td>\n",
       "      <td>0.553317</td>\n",
       "      <td>0.070175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2010-02-10</td>\n",
       "      <td>0.587729</td>\n",
       "      <td>0.704938</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2010-01-13</td>\n",
       "      <td>0.581390</td>\n",
       "      <td>0.703822</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2008-10-20</td>\n",
       "      <td>0.479511</td>\n",
       "      <td>0.228810</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2009-01-22</td>\n",
       "      <td>0.500792</td>\n",
       "      <td>0.290000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2008-01-11</td>\n",
       "      <td>0.415440</td>\n",
       "      <td>0.745338</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2007-05-10</td>\n",
       "      <td>0.359746</td>\n",
       "      <td>0.399248</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2010-03-05</td>\n",
       "      <td>0.592936</td>\n",
       "      <td>0.409460</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2008-07-24</td>\n",
       "      <td>0.459588</td>\n",
       "      <td>0.631610</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2009-12-04</td>\n",
       "      <td>0.572334</td>\n",
       "      <td>0.583747</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2009-02-19</td>\n",
       "      <td>0.507132</td>\n",
       "      <td>0.461386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2007-12-20</td>\n",
       "      <td>0.410460</td>\n",
       "      <td>0.776152</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2009-05-22</td>\n",
       "      <td>0.527960</td>\n",
       "      <td>0.411765</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2009-08-14</td>\n",
       "      <td>0.546978</td>\n",
       "      <td>0.083275</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2010-05-14</td>\n",
       "      <td>0.608784</td>\n",
       "      <td>0.568553</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2009-09-17</td>\n",
       "      <td>0.554675</td>\n",
       "      <td>0.269029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>2007-10-25</td>\n",
       "      <td>0.397781</td>\n",
       "      <td>0.393355</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392923</th>\n",
       "      <td>bury times</td>\n",
       "      <td>2013-12-02</td>\n",
       "      <td>0.902649</td>\n",
       "      <td>0.821862</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392924</th>\n",
       "      <td>bury times</td>\n",
       "      <td>2014-04-11</td>\n",
       "      <td>0.932081</td>\n",
       "      <td>0.838238</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392925</th>\n",
       "      <td>bury times</td>\n",
       "      <td>2013-10-12</td>\n",
       "      <td>0.891103</td>\n",
       "      <td>0.615709</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392926</th>\n",
       "      <td>bury times</td>\n",
       "      <td>2014-04-02</td>\n",
       "      <td>0.930043</td>\n",
       "      <td>0.574966</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392927</th>\n",
       "      <td>bury times</td>\n",
       "      <td>2014-02-12</td>\n",
       "      <td>0.918950</td>\n",
       "      <td>0.546658</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392928</th>\n",
       "      <td>bury times</td>\n",
       "      <td>2014-02-01</td>\n",
       "      <td>0.916459</td>\n",
       "      <td>0.675908</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392929</th>\n",
       "      <td>bury times</td>\n",
       "      <td>2014-01-12</td>\n",
       "      <td>0.911931</td>\n",
       "      <td>0.668461</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392930</th>\n",
       "      <td>bury times</td>\n",
       "      <td>2013-12-02</td>\n",
       "      <td>0.902649</td>\n",
       "      <td>0.835742</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392931</th>\n",
       "      <td>open democracy blog</td>\n",
       "      <td>2007-11-21</td>\n",
       "      <td>0.403894</td>\n",
       "      <td>0.899808</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392932</th>\n",
       "      <td>open democracy blog</td>\n",
       "      <td>2007-11-21</td>\n",
       "      <td>0.403894</td>\n",
       "      <td>0.950902</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392933</th>\n",
       "      <td>open democracy blog</td>\n",
       "      <td>2007-11-21</td>\n",
       "      <td>0.403894</td>\n",
       "      <td>0.759011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392934</th>\n",
       "      <td>open democracy blog</td>\n",
       "      <td>2007-11-22</td>\n",
       "      <td>0.404120</td>\n",
       "      <td>0.706254</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392935</th>\n",
       "      <td>open democracy blog</td>\n",
       "      <td>2007-11-21</td>\n",
       "      <td>0.403894</td>\n",
       "      <td>0.953689</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392936</th>\n",
       "      <td>open democracy blog</td>\n",
       "      <td>2007-11-21</td>\n",
       "      <td>0.403894</td>\n",
       "      <td>0.725183</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392937</th>\n",
       "      <td>open democracy blog</td>\n",
       "      <td>2007-11-21</td>\n",
       "      <td>0.403894</td>\n",
       "      <td>0.815784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392938</th>\n",
       "      <td>open democracy blog</td>\n",
       "      <td>2007-11-21</td>\n",
       "      <td>0.403894</td>\n",
       "      <td>0.887846</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392939</th>\n",
       "      <td>local london</td>\n",
       "      <td>2013-02-09</td>\n",
       "      <td>0.835635</td>\n",
       "      <td>0.761232</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392940</th>\n",
       "      <td>local london</td>\n",
       "      <td>2014-03-06</td>\n",
       "      <td>0.923930</td>\n",
       "      <td>0.236160</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392941</th>\n",
       "      <td>local london</td>\n",
       "      <td>2010-05-09</td>\n",
       "      <td>0.607652</td>\n",
       "      <td>0.958237</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392942</th>\n",
       "      <td>local london</td>\n",
       "      <td>2013-04-07</td>\n",
       "      <td>0.848540</td>\n",
       "      <td>0.889434</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392943</th>\n",
       "      <td>local london</td>\n",
       "      <td>2013-11-08</td>\n",
       "      <td>0.897215</td>\n",
       "      <td>0.922470</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392944</th>\n",
       "      <td>ulster defence league on facebook</td>\n",
       "      <td>2012-01-13</td>\n",
       "      <td>0.746661</td>\n",
       "      <td>0.696320</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392945</th>\n",
       "      <td>the cumberland news</td>\n",
       "      <td>2012-01-28</td>\n",
       "      <td>0.750057</td>\n",
       "      <td>0.923302</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392946</th>\n",
       "      <td>norwich evening news 24</td>\n",
       "      <td>2014-04-24</td>\n",
       "      <td>0.935024</td>\n",
       "      <td>0.684972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392947</th>\n",
       "      <td>norwich evening news 24</td>\n",
       "      <td>2014-02-25</td>\n",
       "      <td>0.921893</td>\n",
       "      <td>0.668280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392948</th>\n",
       "      <td>farmers guardian</td>\n",
       "      <td>2012-10-25</td>\n",
       "      <td>0.811410</td>\n",
       "      <td>0.320878</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392949</th>\n",
       "      <td>ilkeston advertiser</td>\n",
       "      <td>2007-05-31</td>\n",
       "      <td>0.364501</td>\n",
       "      <td>0.504623</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392950</th>\n",
       "      <td>manchester evening news</td>\n",
       "      <td>2007-02-22</td>\n",
       "      <td>0.342314</td>\n",
       "      <td>0.212473</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392951</th>\n",
       "      <td>the cumberland news</td>\n",
       "      <td>2011-06-23</td>\n",
       "      <td>0.700475</td>\n",
       "      <td>0.781450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5392952</th>\n",
       "      <td>local london</td>\n",
       "      <td>2014-06-05</td>\n",
       "      <td>0.944532</td>\n",
       "      <td>0.306089</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5392953 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  src_name        date  norm_date  pred_prob\n",
       "0               beccles and bungay journal  2008-12-22   0.493774   0.447541\n",
       "1               beccles and bungay journal  2010-01-08   0.580258   0.583057\n",
       "2               beccles and bungay journal  2008-06-09   0.449400   0.790210\n",
       "3               beccles and bungay journal  2008-11-03   0.482681   0.809476\n",
       "4               beccles and bungay journal  2009-05-15   0.526375   0.193729\n",
       "5               beccles and bungay journal  2007-05-24   0.362916   0.512569\n",
       "6               beccles and bungay journal  2009-03-28   0.515508   0.020062\n",
       "7               beccles and bungay journal  2010-01-13   0.581390   0.207929\n",
       "8               beccles and bungay journal  2007-07-26   0.377179   0.597845\n",
       "9               beccles and bungay journal  2009-12-16   0.575051   0.382439\n",
       "10              beccles and bungay journal  2010-07-05   0.620557   0.232942\n",
       "11              beccles and bungay journal  2007-11-23   0.404347   0.057177\n",
       "12              beccles and bungay journal  2007-12-08   0.407743   0.683320\n",
       "13              beccles and bungay journal  2009-09-11   0.553317   0.070175\n",
       "14              beccles and bungay journal  2010-02-10   0.587729   0.704938\n",
       "15              beccles and bungay journal  2010-01-13   0.581390   0.703822\n",
       "16              beccles and bungay journal  2008-10-20   0.479511   0.228810\n",
       "17              beccles and bungay journal  2009-01-22   0.500792   0.290000\n",
       "18              beccles and bungay journal  2008-01-11   0.415440   0.745338\n",
       "19              beccles and bungay journal  2007-05-10   0.359746   0.399248\n",
       "20              beccles and bungay journal  2010-03-05   0.592936   0.409460\n",
       "21              beccles and bungay journal  2008-07-24   0.459588   0.631610\n",
       "22              beccles and bungay journal  2009-12-04   0.572334   0.583747\n",
       "23              beccles and bungay journal  2009-02-19   0.507132   0.461386\n",
       "24              beccles and bungay journal  2007-12-20   0.410460   0.776152\n",
       "25              beccles and bungay journal  2009-05-22   0.527960   0.411765\n",
       "26              beccles and bungay journal  2009-08-14   0.546978   0.083275\n",
       "27              beccles and bungay journal  2010-05-14   0.608784   0.568553\n",
       "28              beccles and bungay journal  2009-09-17   0.554675   0.269029\n",
       "29              beccles and bungay journal  2007-10-25   0.397781   0.393355\n",
       "...                                    ...         ...        ...        ...\n",
       "5392923                         bury times  2013-12-02   0.902649   0.821862\n",
       "5392924                         bury times  2014-04-11   0.932081   0.838238\n",
       "5392925                         bury times  2013-10-12   0.891103   0.615709\n",
       "5392926                         bury times  2014-04-02   0.930043   0.574966\n",
       "5392927                         bury times  2014-02-12   0.918950   0.546658\n",
       "5392928                         bury times  2014-02-01   0.916459   0.675908\n",
       "5392929                         bury times  2014-01-12   0.911931   0.668461\n",
       "5392930                         bury times  2013-12-02   0.902649   0.835742\n",
       "5392931                open democracy blog  2007-11-21   0.403894   0.899808\n",
       "5392932                open democracy blog  2007-11-21   0.403894   0.950902\n",
       "5392933                open democracy blog  2007-11-21   0.403894   0.759011\n",
       "5392934                open democracy blog  2007-11-22   0.404120   0.706254\n",
       "5392935                open democracy blog  2007-11-21   0.403894   0.953689\n",
       "5392936                open democracy blog  2007-11-21   0.403894   0.725183\n",
       "5392937                open democracy blog  2007-11-21   0.403894   0.815784\n",
       "5392938                open democracy blog  2007-11-21   0.403894   0.887846\n",
       "5392939                       local london  2013-02-09   0.835635   0.761232\n",
       "5392940                       local london  2014-03-06   0.923930   0.236160\n",
       "5392941                       local london  2010-05-09   0.607652   0.958237\n",
       "5392942                       local london  2013-04-07   0.848540   0.889434\n",
       "5392943                       local london  2013-11-08   0.897215   0.922470\n",
       "5392944  ulster defence league on facebook  2012-01-13   0.746661   0.696320\n",
       "5392945                the cumberland news  2012-01-28   0.750057   0.923302\n",
       "5392946            norwich evening news 24  2014-04-24   0.935024   0.684972\n",
       "5392947            norwich evening news 24  2014-02-25   0.921893   0.668280\n",
       "5392948                   farmers guardian  2012-10-25   0.811410   0.320878\n",
       "5392949                ilkeston advertiser  2007-05-31   0.364501   0.504623\n",
       "5392950            manchester evening news  2007-02-22   0.342314   0.212473\n",
       "5392951                the cumberland news  2011-06-23   0.700475   0.781450\n",
       "5392952                       local london  2014-06-05   0.944532   0.306089\n",
       "\n",
       "[5392953 rows x 4 columns]"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "out_df[['src_name', 'date', 'norm_date', 'pred_prob']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "src_lr_coef = []\n",
    "for s in out_df.src_name.unique():\n",
    "    tmp_df = out_df.loc[out_df.src_name == s, ['date', 'pred_prob']]\n",
    "    tmp_df['norm_date'] = pd.to_datetime(tmp_df['date'], format='%Y-%m-%d').astype(int) / 1000.0\n",
    "    scaler = MinMaxScaler(feature_range=(0, 1))\n",
    "    tmp_df['norm_date'] = scaler.fit_transform(tmp_df[['norm_date']])\n",
    "    lr = linear_model.LinearRegression()\n",
    "    lr.fit(tmp_df[['norm_date']], tmp_df['pred_prob'])\n",
    "    src_lr_coef.append((s, lr.coef_[0]))\n",
    "    #break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>src</th>\n",
       "      <th>coef</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>beccles and bungay journal</td>\n",
       "      <td>0.112351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>southern daily echo</td>\n",
       "      <td>-0.181482</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>berwick today</td>\n",
       "      <td>-0.013459</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>wokingham times</td>\n",
       "      <td>-0.001743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>birmingham mail</td>\n",
       "      <td>0.541070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>lancashire evening post</td>\n",
       "      <td>0.224824</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>belfast telegraph</td>\n",
       "      <td>-0.229581</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>essex chronicle</td>\n",
       "      <td>-0.262354</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>chingford guardian</td>\n",
       "      <td>0.128667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>the daily mail</td>\n",
       "      <td>0.083544</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>cornish guardian</td>\n",
       "      <td>0.050696</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>ilkeston advertiser</td>\n",
       "      <td>-0.589270</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>environmental agency uk</td>\n",
       "      <td>0.022126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>edgware &amp; mill hill times</td>\n",
       "      <td>-0.530479</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>the independent uk</td>\n",
       "      <td>-0.016175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>east lothian courier</td>\n",
       "      <td>-0.006417</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>cheshire online</td>\n",
       "      <td>-0.018797</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>banbury guardian</td>\n",
       "      <td>-0.527389</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>cambridge evening news</td>\n",
       "      <td>0.218048</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>bognor today</td>\n",
       "      <td>-0.228189</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>ballymoney today</td>\n",
       "      <td>0.151224</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>retford guardian</td>\n",
       "      <td>-0.028960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>24dash</td>\n",
       "      <td>0.029871</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>international business times uk</td>\n",
       "      <td>0.101109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>bury times</td>\n",
       "      <td>0.025804</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>driffield today</td>\n",
       "      <td>-0.109552</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>plymouth herald</td>\n",
       "      <td>-0.150336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>reuters alertnet</td>\n",
       "      <td>-0.165198</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>eastern daily press</td>\n",
       "      <td>-0.038038</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>basingstoke gazette</td>\n",
       "      <td>0.165824</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>315</th>\n",
       "      <td>worthing herald</td>\n",
       "      <td>0.245239</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>316</th>\n",
       "      <td>dunmow broadcast</td>\n",
       "      <td>0.056119</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>317</th>\n",
       "      <td>business weekly - uk</td>\n",
       "      <td>0.110405</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>318</th>\n",
       "      <td>world nuclear news</td>\n",
       "      <td>-0.067460</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>319</th>\n",
       "      <td>gazette</td>\n",
       "      <td>0.477325</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>320</th>\n",
       "      <td>insurance insight</td>\n",
       "      <td>0.018871</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>321</th>\n",
       "      <td>the press and journal</td>\n",
       "      <td>-0.038252</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>322</th>\n",
       "      <td>lurgan today</td>\n",
       "      <td>-0.297105</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>323</th>\n",
       "      <td>wirral globe</td>\n",
       "      <td>-0.050745</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>324</th>\n",
       "      <td>borehamwood &amp; elstree times</td>\n",
       "      <td>-0.002704</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>325</th>\n",
       "      <td>st albans observer</td>\n",
       "      <td>-0.030711</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>326</th>\n",
       "      <td>gloucestershire county gazette</td>\n",
       "      <td>0.097583</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>327</th>\n",
       "      <td>ballymena times</td>\n",
       "      <td>0.040240</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>328</th>\n",
       "      <td>bury free press</td>\n",
       "      <td>0.042941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>329</th>\n",
       "      <td>driffield times &amp; post</td>\n",
       "      <td>-0.035819</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>330</th>\n",
       "      <td>cumberland &amp; westmorland herald</td>\n",
       "      <td>0.027562</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>331</th>\n",
       "      <td>hendon &amp; finchley, barnet &amp; potters bar, and edgeware &amp; mills hill times</td>\n",
       "      <td>0.611958</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>332</th>\n",
       "      <td>british foreign &amp; commonwealth office (fco)</td>\n",
       "      <td>-0.041356</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>333</th>\n",
       "      <td>horticulture week</td>\n",
       "      <td>-0.016884</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>334</th>\n",
       "      <td>aberdeen independent</td>\n",
       "      <td>0.008391</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>335</th>\n",
       "      <td>dromore today</td>\n",
       "      <td>0.220983</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>336</th>\n",
       "      <td>berwick advertiser</td>\n",
       "      <td>0.040822</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>337</th>\n",
       "      <td>wanstead and woodford guardian</td>\n",
       "      <td>-0.043234</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>338</th>\n",
       "      <td>wimbledon guardian</td>\n",
       "      <td>0.033735</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>339</th>\n",
       "      <td>woking news and mail</td>\n",
       "      <td>0.012714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>340</th>\n",
       "      <td>hillingdon &amp; uxbridge times</td>\n",
       "      <td>-0.106348</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>341</th>\n",
       "      <td>dromore leader</td>\n",
       "      <td>0.211674</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>342</th>\n",
       "      <td>lionheart [paul ray] blog</td>\n",
       "      <td>-0.183576</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>343</th>\n",
       "      <td>open democracy blog</td>\n",
       "      <td>-0.149778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>344</th>\n",
       "      <td>ulster defence league on facebook</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>345 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                          src  \\\n",
       "0                                                  beccles and bungay journal   \n",
       "1                                                         southern daily echo   \n",
       "2                                                               berwick today   \n",
       "3                                                             wokingham times   \n",
       "4                                                             birmingham mail   \n",
       "5                                                     lancashire evening post   \n",
       "6                                                           belfast telegraph   \n",
       "7                                                             essex chronicle   \n",
       "8                                                          chingford guardian   \n",
       "9                                                              the daily mail   \n",
       "10                                                           cornish guardian   \n",
       "11                                                        ilkeston advertiser   \n",
       "12                                                    environmental agency uk   \n",
       "13                                                  edgware & mill hill times   \n",
       "14                                                         the independent uk   \n",
       "15                                                       east lothian courier   \n",
       "16                                                            cheshire online   \n",
       "17                                                           banbury guardian   \n",
       "18                                                     cambridge evening news   \n",
       "19                                                               bognor today   \n",
       "20                                                           ballymoney today   \n",
       "21                                                           retford guardian   \n",
       "22                                                                     24dash   \n",
       "23                                            international business times uk   \n",
       "24                                                                 bury times   \n",
       "25                                                            driffield today   \n",
       "26                                                            plymouth herald   \n",
       "27                                                           reuters alertnet   \n",
       "28                                                        eastern daily press   \n",
       "29                                                        basingstoke gazette   \n",
       "..                                                                        ...   \n",
       "315                                                           worthing herald   \n",
       "316                                                          dunmow broadcast   \n",
       "317                                                      business weekly - uk   \n",
       "318                                                        world nuclear news   \n",
       "319                                                                   gazette   \n",
       "320                                                         insurance insight   \n",
       "321                                                     the press and journal   \n",
       "322                                                              lurgan today   \n",
       "323                                                              wirral globe   \n",
       "324                                               borehamwood & elstree times   \n",
       "325                                                        st albans observer   \n",
       "326                                            gloucestershire county gazette   \n",
       "327                                                           ballymena times   \n",
       "328                                                           bury free press   \n",
       "329                                                    driffield times & post   \n",
       "330                                           cumberland & westmorland herald   \n",
       "331  hendon & finchley, barnet & potters bar, and edgeware & mills hill times   \n",
       "332                               british foreign & commonwealth office (fco)   \n",
       "333                                                         horticulture week   \n",
       "334                                                      aberdeen independent   \n",
       "335                                                             dromore today   \n",
       "336                                                        berwick advertiser   \n",
       "337                                            wanstead and woodford guardian   \n",
       "338                                                        wimbledon guardian   \n",
       "339                                                      woking news and mail   \n",
       "340                                               hillingdon & uxbridge times   \n",
       "341                                                            dromore leader   \n",
       "342                                                 lionheart [paul ray] blog   \n",
       "343                                                       open democracy blog   \n",
       "344                                         ulster defence league on facebook   \n",
       "\n",
       "         coef  \n",
       "0    0.112351  \n",
       "1   -0.181482  \n",
       "2   -0.013459  \n",
       "3   -0.001743  \n",
       "4    0.541070  \n",
       "5    0.224824  \n",
       "6   -0.229581  \n",
       "7   -0.262354  \n",
       "8    0.128667  \n",
       "9    0.083544  \n",
       "10   0.050696  \n",
       "11  -0.589270  \n",
       "12   0.022126  \n",
       "13  -0.530479  \n",
       "14  -0.016175  \n",
       "15  -0.006417  \n",
       "16  -0.018797  \n",
       "17  -0.527389  \n",
       "18   0.218048  \n",
       "19  -0.228189  \n",
       "20   0.151224  \n",
       "21  -0.028960  \n",
       "22   0.029871  \n",
       "23   0.101109  \n",
       "24   0.025804  \n",
       "25  -0.109552  \n",
       "26  -0.150336  \n",
       "27  -0.165198  \n",
       "28  -0.038038  \n",
       "29   0.165824  \n",
       "..        ...  \n",
       "315  0.245239  \n",
       "316  0.056119  \n",
       "317  0.110405  \n",
       "318 -0.067460  \n",
       "319  0.477325  \n",
       "320  0.018871  \n",
       "321 -0.038252  \n",
       "322 -0.297105  \n",
       "323 -0.050745  \n",
       "324 -0.002704  \n",
       "325 -0.030711  \n",
       "326  0.097583  \n",
       "327  0.040240  \n",
       "328  0.042941  \n",
       "329 -0.035819  \n",
       "330  0.027562  \n",
       "331  0.611958  \n",
       "332 -0.041356  \n",
       "333 -0.016884  \n",
       "334  0.008391  \n",
       "335  0.220983  \n",
       "336  0.040822  \n",
       "337 -0.043234  \n",
       "338  0.033735  \n",
       "339  0.012714  \n",
       "340 -0.106348  \n",
       "341  0.211674  \n",
       "342 -0.183576  \n",
       "343 -0.149778  \n",
       "344  0.000000  \n",
       "\n",
       "[345 rows x 2 columns]"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "src_lr_coef_df = pd.DataFrame(src_lr_coef, columns=['src', 'coef'])\n",
    "src_lr_coef_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>coef</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>345.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.002402</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.200805</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>-0.619581</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>-0.102115</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>-0.008259</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.101957</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>0.642612</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             coef\n",
       "count  345.000000\n",
       "mean     0.002402\n",
       "std      0.200805\n",
       "min     -0.619581\n",
       "25%     -0.102115\n",
       "50%     -0.008259\n",
       "75%      0.101957\n",
       "max      0.642612"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "src_lr_coef_df.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "src_lr_coef_df.to_csv('/opt/notebooks/not_news/tabs/lr_year_pred_prob_coef_by_source.csv', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "sns.set(color_codes=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f9afd612d30>"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "tmp_df = out_df.loc[out_df.src_name == 'berwick today', ['date', 'pred_prob']]\n",
    "tmp_df['norm_date'] = pd.to_datetime(tmp_df['date'], format='%Y-%m-%d').astype(int) / 1000.0\n",
    "scaler = MinMaxScaler(feature_range=(0, 1))\n",
    "tmp_df['norm_date'] = scaler.fit_transform(tmp_df[['norm_date']])\n",
    "plt.figure(figsize=(15, 8))\n",
    "sns.regplot(x='norm_date', y='pred_prob', data=tmp_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "beccles and bungay journal 0.38232364827516613\n"
     ]
    }
   ],
   "source": [
    "for s in out_df.src_name.unique():\n",
    "    tmp_df = out_df.loc[out_df.src_name == s, ['norm_date', 'pred_prob']]\n",
    "    lr = linear_model.LinearRegression()\n",
    "    lr.fit(tmp_df[['norm_date']], tmp_df['pred_prob'])\n",
    "    print(s, lr.coef_[0])\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}