Multiclassification settings¶
These parameters are used only by TBoostClassifier with three or more classes, and only when
the tables are selected by the fold_vote selector (prune_selector="fold_vote"). Setting one
of them away from its default for a regression or binary classification model raises an error
where noted.
A multiclassification model has one set of rating tables per class, and pruning selects one set of effects shared by the classes. See also Pruning settings and Fold vote pruning settings.
multiclass_prune_cv¶
Description¶
The selection regime for multiclassification.
True— K-fold cross-validation, as for regression and binary classification:prune_n_foldsfold models are scored on their held-out folds, voted on withprune_min_stability, gated byprune_drop_zandprune_keep_budget, and checked bymulticlass_prune_guard.False— A single train/select split (seeprune_validation_fraction). This is the earlier regime, kept for comparison: it tends to drop tables on selection noise.
Type
bool
Default value
True
multiclass_prune_guard¶
Description¶
Check the selected set of tables as a whole against the full set on honest objects, and add
dropped tables back while the gap exceeds the tolerance (see
multiclass_prune_guard_floor).
Type
bool
Default value
True
multiclass_prune_guard_floor¶
Description¶
The floor of the guard's tolerance, as a fraction of the improvement of the full model's deviance over the class prior.
The guard adds tables back until the pruned model is within
of the full model on honest objects, where \(z_{dn}\) is
prune_guard_z_dn. The floor is the price a simpler
model is allowed to pay. 0 keeps the standard-error term only.
Type
float
Default value
0.002
multiclass_prune_sel_bags¶
Description¶
The number of bags of the models the selection is trained on (the fold models, or the selection model of the single-split regime). Must be at least 1. Any other value than 1 raises an error for regression and binary classification.
Type
int
Default value
1
prune_validation_fraction¶
Description¶
The fraction of the objects used to select the tables in the single-split regime
(multiclass_prune_cv=False). Any other value than the default raises an error in the other
regimes, and for regression and binary classification.
Type
float
Default value
0.15
prune_se_rule¶
Description¶
The selection rule, in standard errors of the held-out deviance estimate. 0 selects the set with the lowest held-out deviance; larger values (for example, the classic one-standard-error rule, 1.0) prune more aggressively and trade deviance for a smaller model.
Supported only for multiclassification: any other value than 0 raises an error for regression
and binary classification. Use prune_table_budget or
prune_box_budget for a smaller model there.
Type
float
Default value
0.0