def update_fit_params(self, X_train, y_train, eval_set, weights):
output_dim, train_labels = infer_multitask_output(y_train)
for _, y in eval_set:
for task_idx in range(y.shape[1]):
check_output_dim(train_labels[task_idx], y[:, task_idx])
self.output_dim = output_dim
self.classes_ = train_labels
self.target_mapper = [{
class_label: index
for index, class_label in enumerate(classes)
} for classes in self.classes_]
self.preds_mapper = [{
index: class_label
for index, class_label in enumerate(classes)
} for classes in self.classes_]
self.updated_weights = weights
filter_weights(self.updated_weights)
def update_fit_params(
self,
X_train,
y_train,
eval_set,
weights,
):
output_dim, train_labels = infer_output_dim(y_train)
for X, y in eval_set:
check_output_dim(train_labels, y)
self.output_dim = output_dim
self._default_metric = ('auc' if self.output_dim == 2 else 'accuracy')
self.classes_ = train_labels
self.target_mapper = {
class_label: index for index, class_label in enumerate(self.classes_)
}
self.preds_mapper = {
index: class_label for index, class_label in enumerate(self.classes_)
}
self.updated_weights = self.weight_updater(weights)