def IxnEncoderWorker.fit(self, inputs, fx_choice=REG.LGB, **params) [source]
Analyze the feature set and mine interaction rules.
Trains the shallow GB model, mines interaction rules
from the tree structure, gates on predictive power,
and stores the surviving binary encodings.
Parameters
inputs : IxnEncoderInput
Configuration including train_set, num_features,
and cat_features.
and cat_features.
fx_choice : str
The interaction mining backend. Defaults to REG.LGB.
**params : dict
Runtime parameters passed to the internal fitter
(gb_max_leaves, gb_estimators, gb_learning_rate,
gb_min_child_weight, ixn_min_target, ixn_min_share,
ixn_max_corr, gini_cutoff).
(gb_max_leaves, gb_estimators, gb_learning_rate,
gb_min_child_weight, ixn_min_target, ixn_min_share,
ixn_max_corr, gini_cutoff).
Returns
IxnEncoder
The fitted encoder instance.