pycaret

Why we included this project

PyCaret is the AutoML library many data scientists reach for when a modeling task arrives without a framework already chosen. It automates the tedious middle of the pipeline: data preparation, splitting, training and comparing candidate models, tuning the strongest performer, and emitting a real sklearn-compatible Pipeline you can serialize or serve directly. The same experiment API covers classification, regression, clustering, anomaly detection, and time series. The 4.0 rewrite goes further, turning that notebook engine into a self-hosted platform with a FastAPI control plane (model registry, run tracking, monitoring, drift detection, lineage) and a React dashboard that runs locally via docker compose. That makes it useful at two scales at once: a low-ceremony experiment tool for a lone analyst, and a lightweight internal ML platform for a small team that wants model management and serving without standing up heavier infrastructure.