Twitter Bot Detector

Goal Detect Twitter Political Bots in the Wild
Tools Python (sci-kit learn)
Data Source Own Research

Python app based on decision tree implementation to classify Twitter bots and identify political bots. Uses only account attributes (tweets, favorites, frequency of engagement and followers)

It is fairly easy to build and deploy a pretty basic decision tree algorithm in this case. The code algorithm code itself is roughly 30 lines - the flask implementation was probably more challenging.

I've also iterated different approaches (e.g. logistic regression initially seemed accurate enough) but given the smaller size of the sample, decision trees proved to be the most effective.

Visit Github Repo