0.2), int(len(kmeans_best.get_clustered) *
0.8), 100)
]
for train_l in train_l_list:
forest = Forest(kmeans_best.get_clustered, 'clusters', train_l)
forest.train()
forest.predict()
analyzer.make_features_rate(forest.get_feature_importances,
kmeans_best.get_data.columns, train_l)
forest_test_scores.append(
forest.collect_test_score('cross_validate'))
forest_train_scores.append(
forest.collect_train_score('cross_validate'))
analyzer.probability_per_cluster(kmeans_best.get_test)
analyzer.normal_check()
analyzer.calc_predictors_interval()
test_len_list = [
len(dp.get_dataset_no_useless) - i for i in train_l_list
]
if type(forest_test_scores[0]) == numpy.ndarray:
train_val = list(zip(*forest_train_scores))
test_val = list(zip(*forest_test_scores))
for counter, metrics in enumerate(train_val):
vis.make_overfit_check_plot(
train_val[counter], test_val[counter], train_l_list,
f'forest_test/random_forest_for_best_clustering_{counter}_crossval'
)
else:
vis.make_overfit_check_plot(
forest_train_scores, forest_test_scores, train_l_list,
