scikit-opls¶

Orthogonal Projections to Latent Structures (OPLS / OPLS-DA) with a scikit-learn interface.

OPLS (Trygg & Wold, 2002) splits the variation in X into a predictive part correlated with y and orthogonal parts uncorrelated with y, removes the orthogonal variation with an OSC-style orthogonal filter, then fits a standard PLS engine on the cleaned X. With n_orthogonal=0 it reduces exactly to PLSRegression.

Highlights¶

• OPLS — regressor and supervised transformer.
• O2PLS — two-block joint/orthogonal decomposition with bidirectional prediction.
• Cross-validated n_orthogonal selection via scikit-learn's GridSearchCV (see Quickstart).
• OPLSDA — binary classifier composing OPLS.
• Lazy VIP scores on the estimators (OPLS.vip_ / ortho_vip_, also on OPLSDA), usable for feature selection via SelectFromModel(..., importance_getter="vip_").
• Plotting — score and S-plot Displays.
• Validation — permutation significance testing.

All estimators pass scikit-learn's check_estimator compliance suite, support clone / get_params / set_params, and work inside Pipeline and GridSearchCV.

See Installation and Quickstart to get going.