Twitter Sentiment Analysis & Botometer Bot likelihood

July 2020

For this time around I chose to do a project that focus on my brand spanking new Twitter Api. After thought I wanted do a sentiment analysis on the tweets and to classify if a tweet is either positive or negative. I also had another project in mind and decided to combine them and added Botometer. Botometer is a package that can rate the likelihood the Twitter user is a bot or human at a threshold.

What Did I Do?

• Twitter Api

  • First had to sign up for a Twitter Api

  • Hid it in a custom .py file so my credentials wouldn't be revoked or get billed from joyriders...

• The Data

  • Wrote a function that uses tweepy's api call to pull tweets using a 'keyword'; and returns the pulled tweets as a Pandas DataFrame. Calling tweepy I focused on only if the tweet is in english and the tweet must not be a retweet. If they follow these criteria the function will pull the username of the tweeter, the datetime of the tweet and finally the actual tweet in a string format.

  • I Pulled 10,000 tweets on each of two topics. First on Donald Trump and then on a Manhwa (Korean Manga) that I love (but it's on hiatus right now) 'Solo Leveling'. It comes highly recommended if your looking for a great action read.

• Sentiment Analysis on tweets:

  • I used a python package (TextBlob) that will provide me with a nice and easy sentiment polarity score. The polarity score is a float with values ranging from -1 to +1 (100% negative sentiment to 100% positive sentiment, respectably).

  • I wrapped sentiment polarity up in a function for easier calling.

  • Next I preformed the sentiment polarity function on the entire dataframe by using a simply lambda function. Had to censor user's screen name

  • Lastly I classified the sentiment polarity score by positive/negative and displayed the count for each.

• Botometer

  • Lastly I ran Botometer, using Twitter's Api, on a few users. Botometer runs analysis on whether the user is likely a bot or a human.

Hope You enjoyed my A Twitter Sentiment Analysis & Botometer Project

Add me on LinkedIn if you liked.

Please check out my blogs on this project:

Blog - Twitter Sentiment Analysis & Botometer — Part 1: Twitter API

Blog - Twitter Sentiment Analysis & Botometer — Part 2: The Process

Any claps or likes would be very much appreciated.

Built With

• Python 3.8
• Jupyter Notebook 6.0.0
• A few of the main imports: Pandas, Tweepy, Requests, TextBlob, Botometer

Authors

• Samuel Diaz - Creator - sdman135 - LinkedIn