- Tiny implementation of important algorithms in scikit-learn (e.g., pure python, no input validation, no speed/memory optimization, do not support sparse matrix and multioutput).
- Useful when understanding ML algorithms and scikit-learn.
- Multiple implementations of each algorithm.
- Roughly follow the structure of scikit-learn.
- Roughly follow the API standard of scikit-learn.
- Results are compared with scikit-learn.
- calibration (sklearn.calibration)
- cluster (sklearn.cluster)
- covariance (sklearn.covariance)
- decomposition (sklearn.decomposition)
- discriminant_analysis (sklearn.discriminant_analysis)
- dummy (sklearn.dummy)
- ensemble (sklearn.ensemble)
- feature_extraction (sklearn.feature_extraction)
- feature_selection (sklearn.feature_selection)
- impute (sklearn.inpute)
- inspection (sklearn.inspection)
- kernel_ridge (sklearn.kernel_ridge)
- linear_model (sklearn.linear_model)
- metrics (sklearn.metrics) - classification metrics
- metrics (sklearn.metrics) - regression metrics
- metrics (sklearn.metrics) - pairwise metrics
- mixture (sklearn.mixture)
- neighbors (sklearn.neighbors)
- model_selection (sklearn.model_selection) - splitter classes / functions
- model_selection (sklearn.model_selection) - hyper-parameter optimizers
- model_selection (sklearn.model_selection) - model validation
- multiclass (sklearn.multiclass)
- naive_bayes (sklearn.naive_bayes)
- neural_network (sklearn.neural_network)
- preprocessing (sklearn.preprocessing)
- svm (sklearn.svm)
- tree (sklearn.tree)