oadr_cpep.fit module

Phase 2 (site): fit the analytical methods on a given feature set.

One function per process — fit_ridge, fit_lasso, fit_rf — plus fit_models, the convenience that runs all three. Each fits its model on all rows (the coefficient vector / forest that goes to the aggregator), then evaluates the site’s solo performance by 5-fold CV, writes a metrics CSV, prints R²/MSE, and draws its graphic (via plot.py). Every output is stamped with the feature source (from-<src>) so you can see what it was fit on.

oadr_cpep.fit.fit_ridge(site, panel='B', features=None, *, tidy=None, aa=None, demo=None, cpeptide=None, arms=None, arm_subjects=None, outdir='.', alpha=1.0, n_boot=2000, seed=42)

Fit Ridge(alpha) on the feature set -> coefficient vector, CV metrics, graphic.

oadr_cpep.fit.fit_lasso(site, panel='B', features=None, *, tidy=None, aa=None, demo=None, cpeptide=None, arms=None, arm_subjects=None, outdir='.', alpha=0.008, n_boot=2000, seed=42)

Fit Lasso(alpha) on the feature set -> coefficient vector, CV metrics, graphic.

oadr_cpep.fit.fit_rf(site, panel='B', features=None, *, tidy=None, aa=None, demo=None, cpeptide=None, arms=None, arm_subjects=None, outdir='.', n_trees=200, n_boot=2000, seed=42)

Fit a Random Forest on the feature set -> forest pickle, CV metrics, graphic.

oadr_cpep.fit.fit_models(site, panel='B', features=None, *, tidy=None, aa=None, demo=None, cpeptide=None, arms=None, arm_subjects=None, outdir='.', ridge_alpha=1.0, lasso_alpha=0.008, n_trees=200, n_boot=2000, seed=42)

Convenience: run fit_ridge, fit_lasso, fit_rf on the same feature set.