Calculates a "polarity score" and other notable stats on specific Youtube videos. The Polarity Score represents the overall viewer sentiment towards the video on a scale from -1 to 1 with one being the best sentiment and -1 being the worst.
You need python3 and pip installed to get these libraries. Make sure the csv and json modules are installed or included within python's standard library.
Run:
pip install nltk
You'll also need to install vader_lexicon.
Go into a python console and run:
import nltk
nltk.download()
A downloader will pop up. If an SSL Certificate error pops up, navigate to /Applications/Python 3.7 then run Install Certificates.command. Click on the packages tab and scroll down to 'vader_lexicon'. Double click that and once it says that it is installed, exit that downloader and the python console. Install twython, matplotlib, and pandas if you haven't already done so:
pip install twython
pip install matplotlib
pip install pandas
- Go to https://console.developers.google.com/project and sign in to your google account.
- Click on Project -> Create Project
- Select Youtube Data API.
- Click Enable and follow the prompted steps to get your credentials.
To analyze a video, go on Youtube and get the video id found at the end of the url.
For example:
https://www.youtube.com/watch?v=YI3HD0HAz74
yields a video id of YI3HD0HAz74. It's always found at the end of the url after v=
The script will save a CSV file in the local directory once ran. Navigate to the folder where you stored the script and run:
python3 main.py
The script will output the mean polarity score of all the comments and the min, and max polarity score in that set of comments. It also gives a distribution of the number of comments within each polarity score interval.
A CSV file with all the scores and comments will be saved in a CSV file named polarity_[VIDEO ID].csv. You can manipulate those statistics however you like.
Enjoy! Let me know if you find any bugs or had trouble running the script.