I analyzed perspectives, opinions, thoughts of the people engaging in the Brexit tweet trends. The analyses are motivated by the following areas:
- Sentiment polarity of the tweets.
- Category of the tweets.
- Aversion and affection polarity of the tweets.
- Frequency of the words.
- Tagging of the named entities.
- ...
This section depicts some sample end result of each analyses from the whole end results.
The following shows the sentiment polarity of the tweets:
The following table shows the 10 most probable topics with the 10 most probable words in them for the positive tweets using LDA:
The following shows the word map of unpreprocessed positive tweets:
The following shows the 20 most frequent named-entities for positive tweets:
The following shows the 20 most frequent named-entities for neutral tweets:
The following shows the most co-occurring words with Brexit at window size 7 for positive classification:
The following shows 20 most frequent named-entities for neutral category containing the modals like shall, must and need:
- enter the following in the command line:
git clone https://github.com/Cyrus-Rock/twitter-brexit-analysis.git
- you need to download the dataset from here: Twitter Brexit Dataset
- place the downloaded dataset alongside the final_project.ipynb file, you don't need to extract it
- on the command line go to the directory you have cloned the repo
- now enter following on the command line:
jupyter notebook
- on the opened webpage click on the name of the ipynb file (final_project.ipynb)
- now each node could be run individually and the results are shown below each cell
The requirements.txt file, lists all the packages that are required to run the whole project without any problem. You can use the following command to install them in your machine:
python -m pip install -r requirements.txt