durgeshgowdac.Loading
Exploratory analysis hunting for the statistical tells that separate bot accounts from real ones — follower ratios, tweet frequency, and account-creation hours, visualized before any model gets to guess.
Somewhere on Twitter, an account is tweeting every four minutes, following 8,000 people, and followed by roughly none of them. This notebook goes looking for exactly that kind of statistical tell across 37,000+ labeled accounts — account metadata, tweeting rates, creation hours, follower dynamics — visualized to build intuition for what separates a bot from a human before any model gets to make a guess.
Don’t take my word for it — the receipts are on GitHubThe recurring lesson across every EDA project I've done is the same one: the visualization phase is where you actually find the signal, and the modeling step afterward is often just confirming what the plots already showed you.