H2O-3 — AutoML для табличных данных

★ 7.5K

H2O-3 is an open-source distributed machine-learning platform with automated ML (AutoML): it trains and compares many models on your data and returns the best. Reach for it when you need to build a predictive model on tabular data without hand-picking algorithms: churn prediction, scoring, classification, regression — you load data, run AutoML, and get a set of trained models (GBM, random forest, GLM, deep nets, stacking) with metrics and interpretability. It scales across a cluster and has Python, R, and web-UI (Flow) interfaces. Its focus is training and automated selection of ML models on structured data, not BI analytics or working with LLMs: it is a classic ML/AutoML engine for data scientists and analysts. It has its own facilities for serving and explaining models in production; data and training stay on your side.