Modelkit.Lasso_pathA descending lasso regularization path fitted with deterministic warm starts.
Without explicit alphas, fit constructs count logarithmically spaced values from the smallest alpha producing the all-zero centered solution to epsilon times that value. Explicit alphas are copied, validated, and sorted descending. Coefficient-matrix rows, intercepts, reports, and model indices all use this same order. A path of a alpha values costs the sum of its warm-started coordinate-descent sweeps and stores O(a * p) fitted coefficients.
val fit :
t ->
?alphas:Vector.t ->
?sample_weight:Sample_weight.t ->
rng:Rng.t ->
feature_schema:Feature_schema.t ->
x:Matrix.t ->
y:Target.regression Target.t ->
unit ->
(fitted, Error.t) resultval reports : fitted -> Solver_report.t arrayval model : fitted -> index:int -> (Lasso_regression.fitted, Error.t) result