API reference¶
aspective follows scikit-learn's module layout, so each estimator lives where its vanilla version lives in scikit-learn (or, for anomaly detectors, in PyOD).
| Module | Estimators |
|---|---|
aspective.discriminant_analysis |
Regularized and mixture discriminant analysis |
aspective.naive_bayes |
Semi-naive Bayes: AODE, A2DE, TAN, KDB, hidden naive Bayes, flexible naive Bayes |
aspective.neighbors |
Adaptive-metric k-NN (DANN) and LOESS |
aspective.ensemble |
Rotation Forest, LogitBoost, MultiBoost, local linear forests |
aspective.tree |
M5 model trees |
aspective.linear_model |
Firth and Cauchy-prior logistic regression, relaxed, adaptive and nonconvex lasso, Bolasso, L2Boost, MARS, P-spline GAMs |
aspective.anomaly |
RS-Hash, LeSiNN, Extended Isolation Forest |
Every supervised estimator passes scikit-learn's check_estimator, so it
works with pipelines, GridSearchCV, cross_val_score, clone and the rest
of scikit-learn. Every anomaly detector follows PyOD's BaseDetector
contract, and is a PyOD detector when PyOD is installed.