def run_training(self):
self.binary_csp = BinaryCSP(
self.cnt,
self.filterbands,
self.filt_order,
self.folds,
self.class_pairs,
self.segment_ival,
self.n_top_bottom_csp_filters,
standardize_filt_cnt=self.standardize_filt_cnt,
standardize_epo=self.standardize_epo,
ival_optimizer=self.ival_optimizer,
marker_def=self.marker_def)
self.binary_csp.run()
self.filterbank_csp = FilterbankCSP(
self.binary_csp,
n_features=self.n_selected_features,
n_filterbands=self.n_selected_filterbands,
forward_steps=self.forward_steps,
backward_steps=self.backward_steps,
stop_when_no_improvement=self.stop_when_no_improvement)
self.filterbank_csp.run()
self.multi_class = MultiClassWeightedVoting(
self.binary_csp.train_labels_full_fold,
self.binary_csp.test_labels_full_fold,
self.filterbank_csp.train_pred_full_fold,
self.filterbank_csp.test_pred_full_fold, self.class_pairs)
self.multi_class.run()
def recreate_multi_class(train_csp_obj):
""" Assumes filterbank + possibly binary csp was rerun and
recreates multi class weighted voting object
with new labels + predictions. """
train_csp_obj.multi_class = MultiClassWeightedVoting(
train_csp_obj.binary_csp.train_labels_full_fold,
train_csp_obj.binary_csp.test_labels_full_fold,
train_csp_obj.filterbank_csp.train_pred_full_fold,
train_csp_obj.filterbank_csp.test_pred_full_fold,
train_csp_obj.class_pairs)
