IV - Let's find duplicates.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "24809"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "titles = list(Work.objects.filter(category__slug='anime').order_by('title').values_list('title', flat=True))\n",
    "len(titles)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "\n",
    "offset = random.randint(0, len(titles) - 100)\n",
    "#offset = 8975"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Isobe Isobee Monogatari: Ukiyo wa Tsurai yo (2015)\n",
      "Isobe Isobee with Nakata Yasutaka de Sourou\n",
      "Isourou Tengoku\n",
      "I''s Pure\n",
      "I``s Pure\n",
      "I''s Pure Bonus\n",
      "Issei Kiyuukyoku: Divertimento\n",
      "Issho ni Ecchi\n",
      "Issho ni H Shiyo\n",
      "Issho ni Sleeping: Sleeping with Hinako\n",
      "Issho ni Training Ofuro: Bathtime with Hinako & Hiyoko\n",
      "Issho ni Training: Training with Hinako\n",
      "Isshuukan Friends. Kaori no Nikki\n",
      "Isshuukan Friends. Specials\n",
      "Isshuukan Friends.: Tomodachi to no Omoide\n",
      "Issunboushi\n",
      "Issunboushi (1956)\n",
      "Issunboushi (2007)\n",
      "Issunboushi: Chibi Jo Monogatari\n",
      "Is the order a rabbit?\n"
     ]
    }
   ],
   "source": [
    "subset = titles[offset:offset + 20]\n",
    "for title in subset:\n",
    "    print(title)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "\n",
    "vec = CountVectorizer(analyzer='char', ngram_range=(2, 2))\n",
    "X = vec.fit_transform(subset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 6.75 ms, sys: 514 µs, total: 7.27 ms\n",
      "Wall time: 4.74 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "\n",
    "sim = cosine_similarity(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.imshow(sim)\n",
    "plt.colorbar()\n",
    "t = np.arange(len(subset))\n",
    "plt.yticks(t, subset)\n",
    "plt.xticks(t, [])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Artist objects with same name or first name/last name inversed"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 occurrence : 2445 artiste(s)\n",
      "2 occurrence : 83 artiste(s)\n",
      "3 occurrence : 6 artiste(s)\n",
      "4 occurrence : 2 artiste(s)\n",
      "\n",
      "kawahara reki : 4 occurrences\n",
      "harold sakuishi : 4 occurrences\n",
      "araki hirohiko : 3 occurrences\n",
      "otonaka sawaki : 3 occurrences\n",
      "akira toriyama : 3 occurrences\n",
      "miki yoshikawa : 3 occurrences\n",
      "sukeno yoshiaki : 3 occurrences\n",
      "ai yazawa : 3 occurrences\n"
     ]
    }
   ],
   "source": [
    "from collections import Counter\n",
    "\n",
    "c = Counter()\n",
    "for name in Artist.objects.values_list('name', flat=True):\n",
    "    c[' '.join(sorted(name.lower().split()))] += 1\n",
    "meta = Counter(c.values())\n",
    "for k, v in meta.most_common():\n",
    "    print(k, 'occurrence :', v, 'artiste(s)')\n",
    "print()\n",
    "for k, v in c.most_common():\n",
    "    if v >= 3:\n",
    "        print(k, ':', v, 'occurrences')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## How many works share a title"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('The Asterisk War: The Academy City on the Water', 'anime') 9\n",
      "('Dororo', 'anime') 4\n",
      "('Zatch Bell!', 'anime') 4\n",
      "('Yakusoku', 'anime') 4\n",
      "('Sorcerer Hunters', 'anime') 3\n",
      "('Naita Aka Oni', 'anime') 3\n",
      "('Meat or Die', 'anime') 3\n",
      "('Saru Kani Gassen', 'anime') 3\n",
      "('Shiroi Zou', 'anime') 3\n",
      "('Meow no Hoshi', 'anime') 3\n",
      "('Those Who Hunt Elves', 'anime') 3\n",
      "('Digimon: The Movie', 'anime') 3\n",
      "('Yu-Gi-Oh!', 'anime') 3\n",
      "('Mottainai', 'anime') 3\n",
      "('Mikagura Gakuen Kumikyoku', 'anime') 3\n",
      "('Dragon Quest: Dai no Daibouken', 'anime') 3\n",
      "('Street Fighter Zero: The Animation', 'anime') 3\n",
      "('Desert Punk', 'anime') 3\n",
      "('Imawa no Kuni no Alice', 'anime') 3\n",
      "('DRAMAtical Murder OVA: Data_xx_Transitory', 'anime') 3\n",
      "('Nekopara', 'anime') 3\n",
      "('Black Clover', 'anime') 3\n",
      "('Fastening Days', 'anime') 3\n",
      "('Shiranpuri', 'anime') 3\n",
      "('Kekkaishi', 'anime') 3\n",
      "('Hunter x Hunter', 'anime') 3\n",
      "('Dimension W', 'anime') 3\n",
      "('Shinpi no Sekai El Hazard', 'anime') 3\n",
      "('Mahou Tsukai Tai!', 'anime') 3\n",
      "('Yamada-kun to 7-nin no Majo', 'anime') 3\n",
      "('Doukyuusei', 'anime') 3\n",
      "('Buddy Go!', 'anime') 3\n",
      "('Eiga Daisuki Pompo-san', 'anime') 3\n",
      "('Feng Ling Yu Xiu', 'anime') 3\n",
      "('Sakkai Eightraid', 'anime') 3\n",
      "('Shadowverse', 'anime') 3\n"
     ]
    }
   ],
   "source": [
    "c = Counter(Work.objects.values_list('title', 'category__slug'))\n",
    "for k, v in c.most_common():\n",
    "    if v >= 3:  # If 2 then there are too many\n",
    "        print(k, v)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### What about alternative titles?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "nb = 0\n",
    "alt_titles = dict(WorkTitle.objects.values_list('title', 'work_id'))\n",
    "pairs = []\n",
    "for work_id, title in Work.objects.filter(title__in=alt_titles.keys()).values_list('id', 'title'):\n",
    "    if work_id != alt_titles[title]:\n",
    "        pairs.append((work_id, alt_titles[title]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "works = Work.objects.in_bulk(list(np.array(pairs).flatten()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1479"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(pairs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "21458 Lupin the IIIrd: Chikemuri no Ishikawa Goemon / Lupin III: Goemon Ishikawa's Spray of Blood 28327\n",
      "233 Doraemon / Doraemon (1979) 11292\n",
      "13471 Haikyuu!!: Jump Festa 2015 Special / Haikyuu!!: vs. \"Akaten\" 25374\n",
      "13827 Pandora in the Crimson Shell: Ghost Urn / Koukaku no Pandora: Ghost Urn 27797\n",
      "2951 I love BuBu ChaCha / Daisuki! Bubu- Chacha 23460\n",
      "2983 UFO Ultramaiden Valkyrie 4: Banquet of Time, Dreams, and Galaxies / UFO Princess Valkyrie: Toki to Yume to Ginga no Utage 34014\n",
      "1177 Porong Porong Pororo / Bbororoe Daemoheom 22450\n",
      "3254 Great Yamato No. 0 / Dai Yamato Zero-go 23443\n",
      "3286 Fly, Space Battleship Geobukseon / Narara Ujujeonham Geobukseon 29626\n",
      "2118 Voltage Fighter Gowcaizer / Choujin Gakuen Gowcaizer 23159\n"
     ]
    }
   ],
   "source": [
    "for work_id, work_id2 in pairs[:10]:\n",
    "    if works[work_id].title != works[work_id2].title:\n",
    "        print(work_id, works[work_id].title, '/', works[work_id2].title, work_id2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## How many works share a reference?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from django.db.models import Count\n",
    "\n",
    "def describe(queryset):\n",
    "    nb_distinct_fields = queryset.filter(nb__gte=2).count()\n",
    "    nb_duplicates = sum(queryset.filter(nb__gte=2).values_list('nb', flat=True))\n",
    "    print('{:d} oeuvres partagent {:d} champs'.format(nb_duplicates, nb_distinct_fields))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "575 oeuvres partagent 281 champs\n"
     ]
    }
   ],
   "source": [
    "queryset = Work.objects.values('reference__url').annotate(nb=Count('reference__url')).order_by('-nb')\n",
    "describe(queryset)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Donc en septembre 2018, 577 œuvres partagent 282 références. Peut-être encore des doublons à écarter."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## How many works have same poster?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<WorkQuerySet [{'ext_poster': 'https://cdn.anidb.net/images/main/165165.jpg', 'nb': 3}, {'ext_poster': 'https://s4.anilist.co/file/anilistcdn/media/anime/cover/medium/default.jpg', 'nb': 3}, {'ext_poster': 'https://cdn.anidb.net/images/main/231328.jpg', 'nb': 3}, {'ext_poster': 'http://www.manga-news.com/public/images/series/dragonhuntercoffret1gd.jpg', 'nb': 2}, {'ext_poster': 'https://cdn.anidb.net/images/main/11295.jpg', 'nb': 2}]>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "queryset = Work.objects.filter(ext_poster__endswith='.jpg').values('ext_poster').annotate(nb=Count('ext_poster')).order_by('-nb')\n",
    "queryset[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "99 oeuvres partagent 48 champs\n"
     ]
    }
   ],
   "source": [
    "describe(queryset)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## How many works have same AniDB ID?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<WorkQuerySet [{'anidb_aid': 4932, 'nb': 6}, {'anidb_aid': 8778, 'nb': 3}, {'anidb_aid': 6747, 'nb': 2}, {'anidb_aid': 4897, 'nb': 2}, {'anidb_aid': 9977, 'nb': 2}]>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "queryset = Work.objects.exclude(anidb_aid=0).values('anidb_aid').annotate(nb=Count('anidb_aid')).order_by('-nb')\n",
    "queryset[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35 oeuvres partagent 15 champs\n"
     ]
    }
   ],
   "source": [
    "describe(queryset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<WorkQuerySet [<Work: Kara no Kyoukai 3: Tsuukaku Zanryuu>, <Work: Kara no Kyoukai 1: Fukan Fuukei>, <Work: Kara no Kyoukai 2: Satsujin Kousatsu (Part 1)>, <Work: Kara no Kyoukai 5: Mujun Rasen>, <Work: Kara no Kyoukai 4: Garan no Dou>, <Work: Kara no Kyoukai 7: Satsujin Kousatsu (Part 2)>]>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Work.objects.filter(anidb_aid=4932)  # Lol ce n'est pas un doublon"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<WorkQuerySet [<Work: Puella Magi Madoka Magica the Movie Part 3: Rebellion>, <Work: Puella Magi Madoka Magica the Movie Part 1: Beginnings>, <Work: Puella Magi Madoka Magica the Movie Part 2: Eternal>]>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Work.objects.filter(anidb_aid=8778)  # Non plus"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## How many anime have an AniDB ID?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5785"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nb_anidb_works = Work.objects.exclude(anidb_aid=0).count()\n",
    "nb_anidb_works"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.23318150671127413"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nb_anidb_works / Work.objects.filter(category__slug='anime').count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## How many works have at least one Reference?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8687271858902194"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Work.objects.annotate(nb=Count('reference')).values('id').filter(nb__gte=1).count() / Work.objects.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "anime\n",
      "24788/24809 = 99.9 %\n",
      "manga\n",
      "677/4503 = 15.0 %\n",
      "album\n",
      "0/1 = 0.0 %\n"
     ]
    }
   ],
   "source": [
    "for category in Category.objects.all():\n",
    "    print(category.slug)\n",
    "    category_works = Work.objects.filter(category=category)\n",
    "    nb = category_works.annotate(nb=Count('reference')).values('id').filter(nb__gte=1).count()\n",
    "    print('{:d}/{:d} = {:.1f} %'.format(nb, category_works.count(), 100 * nb / category_works.count()))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Django Shell-Plus",
   "language": "python",
   "name": "django_extensions"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}