scikit-learn /

scikit-learn

Live map
View repository
Snapshotmain · cedc9d8

How estimator contracts compose preprocessing, fitting, cross-validation, scoring, and model selection.

Click a building or route
Choose a path

Transform selected columns, pass the resulting features through Pipeline, and fit the predictor.

System map7 components · 7 connections
130%
scikit-learn architectureData enters through splitters and feature composition on the left, the estimator sits at center, and validation plus scoring fan out across workers on the right.transformers and colum…concatenated transform…X transformed, yColumnTransformerEstimatorcontractConcreteestimatorPipelineCV splitterValidationrunnerScorer
Connections
BuildingsEntryServiceComputeStorageRuntimeToolingExternal

Terms, in plain English

EstimatorAn object whose constructor parameters describe configuration and whose fit method learns state.

How to study this repo

  1. 1
    Inspect the estimator protocol

    Read base.py, then find how Pipeline exposes and clones nested step parameters.

  2. 2
    Follow one CV fold

    Trace indices from KFold through _fit_and_score, Pipeline.fit, prediction, and the scorer.