[snikolov]

A nonparametric model for anomaly detection, classification, and prediction in timeseries. It observes the trajectory of some measurement over time and compares it to tons of historical trajectories to see if it looks like anything that has happened before. Based on what it looks like, you can predict where it will go next, classify it (if the historical trajectories are labeled), or detect anomalies (if it doesn't look sufficiently like anything seen so far).