model = ModelHMM() model.train(data_ref, labels_ref, list_features_final, dim_features) data_ref = [] for data in data_test: df = pd.DataFrame(data) df.columns = list_features data_ref.append(df[list_features_final].values) pred_labels, proba = model.test_model(data_ref) F1_temp = [] for i in range(len(labels_test)): F1_temp.append( tools.compute_F1_score(labels_test[i], pred_labels[i], list_states[num_track])) F1_score.append(np.mean(F1_temp)) dim_score.append(dim) feaures_save.append(str(list_features_final)) score_totaux = pd.DataFrame({ 'nbr_components': dim_score, 'score': F1_score, 'features': feaures_save }) score_totaux.to_csv('score/score_model_wrapper_' + name_track + "2.csv", index=False) dim += 1
time_test = [] for id_subject in id_test: time_test.append(timestamps[id_subject]) for i in range(len(labels_test)): conf_mat = tools.compute_confusion_matrix( predict_labels_fisher[i], labels_test[i], list_states) confusion_matrix_fisher += conf_mat conf_mat = tools.compute_confusion_matrix( predict_labels_wrapper[i], labels_test[i], list_states) confusion_matrix_wrapper += conf_mat F1_fisher.append( tools.compute_F1_score(labels_test[i], predict_labels_fisher[i], list_states)) F1_wrapper.append( tools.compute_F1_score(labels_test[i], predict_labels_wrapper[i], list_states)) total += 1 F1_f.append(np.mean(F1_fisher)) F1_w.append(np.mean(F1_wrapper)) F1_fisher_temp += np.mean(F1_fisher) F1_wrapper_temp += np.mean(F1_wrapper) index_nbr_f.append(nbr_features - 1) index_nbr_f.append(nbr_features - 1)
for data in data_ref_all[nbr_test]: df_data = pd.DataFrame(data, columns = list_features) data_ref.append(df_data[sub_list_features].values) for data in data_test_all[nbr_test]: df_data = pd.DataFrame(data, columns = list_features) data_test.append(df_data[sub_list_features].values) model = ModelHMM() model.train(data_ref, labels_ref[num_track][nbr_test], sub_list_features, np.ones(len(sub_list_features))) #### Test predict_labels, proba = model.test_model(data_test) for i in range(len(predict_labels)): F1_score.append(tools.compute_F1_score(labels_test[num_track][nbr_test][i], predict_labels[i], list_states[num_track])) F1_S[num_track] = np.mean(F1_score) score_total[num_track].append(F1_S[num_track]) score_totaux = pd.DataFrame( {'best_features': best_features_total}) for name_track, num_track in zip(tracks, range(len(tracks))): df_track = pd.DataFrame( { name_track: score_total[num_track]}) score_totaux = pd.concat([score_totaux, df_track], axis=1) score_totaux = ranking_features(score_totaux, tracks, method_sort) score_totaux.to_csv(path_save + '/' + file_name + str(iteration+1), index=False)
# ax.grid() for id_train in range(len(data_ref)): data_ref[id_train] = pca.transform(data_ref[id_train]) for id_test in range(len(data_test)): data_test[id_test] = pca.transform(data_test[id_test]) model = ModelHMM() model.train(data_ref, labels_ref, col, dim_features) pred_labels, proba = model.test_model(data_test) F1_temp = [] for i in range(len(labels_test)): F1_temp.append(tools.compute_F1_score(labels_test[i], pred_labels[i], list_states)) F1_score.append(np.mean(F1_temp)) dim_score.append(n_components) score_totaux = pd.DataFrame( {'nbr_components': dim_score, 'score': F1_score, }) score_totaux.to_csv('score_pca' + '_' + name_track + ".csv", index=False)
list_states_posture_final = sorted( list_states_posture_final) for i in range(len(predict_labels)): MCC_combined.append( tools.compute_MCC_score( labels_test_detailed_posture[i], labels_final[i], list_states_posture_final)) MCC_detailed.append( tools.compute_MCC_score( labels_test_detailed_posture[i], predict_labels_detailed_posture[i], list_states_posture_final)) MCC_general.append( tools.compute_F1_score(labels_test[i], predict_labels[i], list_states_posture)) MCC_details.append( tools.compute_F1_score(labels_test_details[i], predict_labels_details[i], list_states_details)) F1_combined.append( tools.compute_F1_score(labels_test_detailed_posture[i], labels_final[i], list_states_posture_final)) F1_detailed.append( tools.compute_F1_score( labels_test_detailed_posture[i], predict_labels_detailed_posture[i], list_states_posture_final))