Anomaly Detection — Tiktok

Uses Isolation Forest to detect statistically anomalous profiles — works even without labeled training data.

Back
How does Anomaly Detection work?
Isolation Forest identifies profiles that are statistically unusual — e.g., accounts with extreme follower/following ratios, suspicious bio patterns, or unusual posting frequencies. Unlike the main classifier, no training labels are needed. Higher anomaly score = more suspicious.

Upload Profile CSV

No label column needed — anomaly detection is unsupervised.