The original title is "The Random Forest Cheat Sheet: Everything on One Page"

Short summary

A comprehensive one-page cheat sheet for Random Forest covering the three randomness knobs (bootstrap, max_features, random splits), their measured correlation impact, and starting hyperparameters split by classification vs regression. Includes guidance on importance method selection, a Bagging vs RF vs Extra Trees comparison table with measured accuracy numbers, and OOB validation explained in one block.

• •max_features is the only knob that moves correlation; optimal value differs by problem type
• •Permutation importance on held-out rows is preferred over MDI which rewards high-cardinality columns
• •Extra Trees achieves lowest correlation (0.515) and highest ensemble accuracy (0.9050) at the cost of single-tree accuracy

Generated with AI, which can make mistakes.