def post_evaluation_multimodal(metrics, file, filename, list_of_demonstrations,
feat_fname):
mutual_information_1 = []
normalized_mutual_information_1 = []
adjusted_mutual_information_1 = []
homogeneity_1 = []
precision_1_micro = []
recall_1_micro = []
precision_1_macro = []
recall_1_macro = []
precision_1_weighted = []
recall_1_weighted = []
mutual_information_2 = []
normalized_mutual_information_2 = []
adjusted_mutual_information_2 = []
homogeneity_2 = []
precision_2_micro = []
recall_2_micro = []
precision_2_macro = []
recall_2_macro = []
precision_2_weighted = []
recall_2_weighted = []
silhoutte_level_1_global = []
silhoutte_level_1_weighted = []
dunn1_level_1 = []
dunn2_level_1 = []
dunn3_level_1 = []
silhoutte_level_2_global = []
silhoutte_level_2_weighted = []
dunn1_level_2 = []
dunn2_level_2 = []
dunn3_level_2 = []
list_of_frms = {}
for elem in metrics:
precision_1_micro.append(elem[0]["precision_micro"])
precision_2_micro.append(elem[1]["precision_micro"])
precision_1_macro.append(elem[0]["precision_macro"])
precision_2_macro.append(elem[1]["precision_macro"])
precision_1_weighted.append(elem[0]["precision_weighted"])
precision_2_weighted.append(elem[1]["precision_weighted"])
recall_1_micro.append(elem[0]["recall_micro"])
recall_2_micro.append(elem[1]["recall_micro"])
recall_1_macro.append(elem[0]["recall_macro"])
recall_2_macro.append(elem[1]["recall_macro"])
recall_1_weighted.append(elem[0]["recall_weighted"])
recall_2_weighted.append(elem[1]["recall_weighted"])
mutual_information_1.append(elem[0]["mutual_info_score"])
mutual_information_2.append(elem[1]["mutual_info_score"])
normalized_mutual_information_1.append(
elem[0]["normalized_mutual_info_score"])
normalized_mutual_information_2.append(
elem[1]["normalized_mutual_info_score"])
adjusted_mutual_information_1.append(
elem[0]["adjusted_mutual_info_score"])
adjusted_mutual_information_2.append(
elem[1]["adjusted_mutual_info_score"])
homogeneity_1.append(elem[0]["homogeneity_score"])
homogeneity_2.append(elem[1]["homogeneity_score"])
silhoutte_level_1_global.append(elem[2]["level1"])
dunn1_level_1.append(elem[3]["level1"])
dunn2_level_1.append(elem[4]["level1"])
dunn3_level_1.append(elem[5]["level1"])
silhoutte_level_2_global += elem[6]
dunn1_level_2 += elem[7]
dunn2_level_2 += elem[8]
dunn3_level_2 += elem[9]
viz = elem[10]
for demonstration in viz.keys():
utils.dict_insert_list(demonstration, viz[demonstration],
list_of_frms)
silhoutte_level_1_weighted.append(elem[11]["level1"])
silhoutte_level_2_weighted += elem[12]
utils.print_and_write_2("precision_1_micro", np.mean(precision_1_micro),
np.std(precision_1_micro), file)
utils.print_and_write_2("precision_2_micro", np.mean(precision_2_micro),
np.std(precision_2_micro), file)
utils.print_and_write_2("precision_1_macro", np.mean(precision_1_macro),
np.std(precision_1_macro), file)
utils.print_and_write_2("precision_2_macro", np.mean(precision_2_macro),
np.std(precision_2_macro), file)
utils.print_and_write_2("precision_1_weighted",
np.mean(precision_1_weighted),
np.std(precision_1_weighted), file)
utils.print_and_write_2("precision_2_weighted",
np.mean(precision_2_weighted),
np.std(precision_2_weighted), file)
utils.print_and_write_2("recall_1_micro", np.mean(recall_1_micro),
np.std(recall_1_micro), file)
utils.print_and_write_2("recall_2_micro", np.mean(recall_2_micro),
np.std(recall_2_micro), file)
utils.print_and_write_2("recall_1_macro", np.mean(recall_1_macro),
np.std(recall_1_macro), file)
utils.print_and_write_2("recall_2_macro", np.mean(recall_2_macro),
np.std(recall_2_macro), file)
utils.print_and_write_2("recall_1_weighted", np.mean(recall_1_weighted),
np.std(recall_1_weighted), file)
utils.print_and_write_2("recall_2_weighted", np.mean(recall_2_weighted),
np.std(recall_2_weighted), file)
utils.print_and_write_2("mutual_info_1", np.mean(mutual_information_1),
np.std(mutual_information_1), file)
utils.print_and_write_2("mutual_info_2", np.mean(mutual_information_2),
np.std(mutual_information_2), file)
utils.print_and_write_2("normalized_mutual_info_1",
np.mean(normalized_mutual_information_1),
np.std(normalized_mutual_information_1), file)
utils.print_and_write_2("normalized_mutual_info_2",
np.mean(normalized_mutual_information_2),
np.std(normalized_mutual_information_2), file)
utils.print_and_write_2("adjusted_mutual_info_1",
np.mean(adjusted_mutual_information_1),
np.std(adjusted_mutual_information_1), file)
utils.print_and_write_2("adjusted_mutual_info_2",
np.mean(adjusted_mutual_information_2),
np.std(adjusted_mutual_information_2), file)
utils.print_and_write_2("homogeneity_1", np.mean(homogeneity_1),
np.std(homogeneity_1), file)
utils.print_and_write_2("homogeneity_2", np.mean(homogeneity_2),
np.std(homogeneity_2), file)
utils.print_and_write_2("silhoutte_level_1_global",
np.mean(silhoutte_level_1_global),
np.std(silhoutte_level_1_global), file)
utils.print_and_write_2("silhoutte_level_2_global",
np.mean(silhoutte_level_2_global),
np.std(silhoutte_level_2_global), file)
utils.print_and_write_2("silhoutte_level_1_weighted",
np.mean(silhoutte_level_1_weighted),
np.std(silhoutte_level_1_weighted), file)
utils.print_and_write_2("silhoutte_level_2_weighted",
np.mean(silhoutte_level_2_weighted),
np.std(silhoutte_level_2_weighted), file)
utils.print_and_write_2("dunn1_level1", np.mean(dunn1_level_1),
np.std(dunn1_level_1), file)
utils.print_and_write_2("dunn2_level1", np.mean(dunn2_level_1),
np.std(dunn2_level_1), file)
utils.print_and_write_2("dunn3_level1", np.mean(dunn3_level_1),
np.std(dunn3_level_1), file)
utils.print_and_write_2("dunn1_level2", np.mean(dunn1_level_2),
np.std(dunn1_level_2), file)
utils.print_and_write_2("dunn2_level2", np.mean(dunn2_level_2),
np.std(dunn2_level_2), file)
utils.print_and_write_2("dunn3_level2", np.mean(dunn3_level_2),
np.std(dunn3_level_2), file)
list_of_dtw_values = []
list_of_norm_dtw_values = []
list_of_lengths = []
data_cache = {}
for demonstration in list_of_demonstrations:
try:
list_of_frms_demonstration = list_of_frms[demonstration]
data = {}
for i in range(len(list_of_frms_demonstration)):
data[i] = [elem[0] for elem in list_of_frms_demonstration[i]]
dtw_score, normalized_dtw_score, length, labels_manual_d, colors_manual_d, labels_automatic_d, colors_automatic_d = broken_barh.plot_broken_barh(
demonstration, data, constants.PATH_TO_CLUSTERING_RESULTS +
demonstration + "_" + filename + ".jpg",
constants.N_COMPONENTS_TIME_ZW)
list_of_dtw_values.append(dtw_score)
list_of_norm_dtw_values.append(normalized_dtw_score)
list_of_lengths.append(length)
# Inserting manual and annotations labels into data struct before dumping as pickle file
cache_entry = {}
cache_entry['changepoints'] = list_of_frms_demonstration
cache_entry['plot_labels_manual'] = labels_manual_d
cache_entry['plot_colors_manual'] = colors_manual_d
cache_entry['plot_labels_automatic'] = labels_automatic_d
cache_entry['plot_colors_automatic'] = colors_automatic_d
data_cache[demonstration] = cache_entry
except:
print demonstration
continue
utils.print_and_write_2("dtw_score", np.mean(list_of_dtw_values),
np.std(list_of_dtw_values), file)
utils.print_and_write_2("dtw_score_normalized",
np.mean(list_of_norm_dtw_values),
np.std(list_of_norm_dtw_values), file)
utils.print_and_write(str(list_of_lengths), file)
pickle.dump(
data_cache,
open(constants.PATH_TO_CLUSTERING_RESULTS + filename + "_.p", "wb"))
def post_evaluation_multimodal(metrics, file, filename, list_of_demonstrations, feat_fname):
mutual_information_1 = []
normalized_mutual_information_1 = []
adjusted_mutual_information_1 = []
homogeneity_1 = []
precision_1_micro = []
recall_1_micro = []
precision_1_macro = []
recall_1_macro = []
precision_1_weighted = []
recall_1_weighted = []
mutual_information_2 = []
normalized_mutual_information_2 = []
adjusted_mutual_information_2 = []
homogeneity_2 = []
precision_2_micro = []
recall_2_micro = []
precision_2_macro = []
recall_2_macro = []
precision_2_weighted = []
recall_2_weighted = []
silhoutte_level_1_global = []
silhoutte_level_1_weighted = []
dunn1_level_1 = []
dunn2_level_1 = []
dunn3_level_1 = []
silhoutte_level_2_global = []
silhoutte_level_2_weighted = []
dunn1_level_2 = []
dunn2_level_2 = []
dunn3_level_2 = []
list_of_frms = {}
for elem in metrics:
precision_1_micro.append(elem[0]["precision_micro"])
precision_2_micro.append(elem[1]["precision_micro"])
precision_1_macro.append(elem[0]["precision_macro"])
precision_2_macro.append(elem[1]["precision_macro"])
precision_1_weighted.append(elem[0]["precision_weighted"])
precision_2_weighted.append(elem[1]["precision_weighted"])
recall_1_micro.append(elem[0]["recall_micro"])
recall_2_micro.append(elem[1]["recall_micro"])
recall_1_macro.append(elem[0]["recall_macro"])
recall_2_macro.append(elem[1]["recall_macro"])
recall_1_weighted.append(elem[0]["recall_weighted"])
recall_2_weighted.append(elem[1]["recall_weighted"])
mutual_information_1.append(elem[0]["mutual_info_score"])
mutual_information_2.append(elem[1]["mutual_info_score"])
normalized_mutual_information_1.append(elem[0]["normalized_mutual_info_score"])
normalized_mutual_information_2.append(elem[1]["normalized_mutual_info_score"])
adjusted_mutual_information_1.append(elem[0]["adjusted_mutual_info_score"])
adjusted_mutual_information_2.append(elem[1]["adjusted_mutual_info_score"])
homogeneity_1.append(elem[0]["homogeneity_score"])
homogeneity_2.append(elem[1]["homogeneity_score"])
silhoutte_level_1_global.append(elem[2]["level1"])
dunn1_level_1.append(elem[3]["level1"])
dunn2_level_1.append(elem[4]["level1"])
dunn3_level_1.append(elem[5]["level1"])
silhoutte_level_2_global += elem[6]
dunn1_level_2 += elem[7]
dunn2_level_2 += elem[8]
dunn3_level_2 += elem[9]
viz = elem[10]
for demonstration in viz.keys():
utils.dict_insert_list(demonstration, viz[demonstration], list_of_frms)
silhoutte_level_1_weighted.append(elem[11]["level1"])
silhoutte_level_2_weighted += elem[12]
utils.print_and_write_2("precision_1_micro", np.mean(precision_1_micro), np.std(precision_1_micro), file)
utils.print_and_write_2("precision_2_micro", np.mean(precision_2_micro), np.std(precision_2_micro), file)
utils.print_and_write_2("precision_1_macro", np.mean(precision_1_macro), np.std(precision_1_macro), file)
utils.print_and_write_2("precision_2_macro", np.mean(precision_2_macro), np.std(precision_2_macro), file)
utils.print_and_write_2("precision_1_weighted", np.mean(precision_1_weighted), np.std(precision_1_weighted), file)
utils.print_and_write_2("precision_2_weighted", np.mean(precision_2_weighted), np.std(precision_2_weighted), file)
utils.print_and_write_2("recall_1_micro", np.mean(recall_1_micro), np.std(recall_1_micro), file)
utils.print_and_write_2("recall_2_micro", np.mean(recall_2_micro), np.std(recall_2_micro), file)
utils.print_and_write_2("recall_1_macro", np.mean(recall_1_macro), np.std(recall_1_macro), file)
utils.print_and_write_2("recall_2_macro", np.mean(recall_2_macro), np.std(recall_2_macro), file)
utils.print_and_write_2("recall_1_weighted", np.mean(recall_1_weighted), np.std(recall_1_weighted), file)
utils.print_and_write_2("recall_2_weighted", np.mean(recall_2_weighted), np.std(recall_2_weighted), file)
utils.print_and_write_2("mutual_info_1", np.mean(mutual_information_1), np.std(mutual_information_1), file)
utils.print_and_write_2("mutual_info_2", np.mean(mutual_information_2), np.std(mutual_information_2), file)
utils.print_and_write_2("normalized_mutual_info_1", np.mean(normalized_mutual_information_1), np.std(normalized_mutual_information_1), file)
utils.print_and_write_2("normalized_mutual_info_2", np.mean(normalized_mutual_information_2), np.std(normalized_mutual_information_2), file)
utils.print_and_write_2("adjusted_mutual_info_1", np.mean(adjusted_mutual_information_1), np.std(adjusted_mutual_information_1), file)
utils.print_and_write_2("adjusted_mutual_info_2", np.mean(adjusted_mutual_information_2), np.std(adjusted_mutual_information_2), file)
utils.print_and_write_2("homogeneity_1", np.mean(homogeneity_1), np.std(homogeneity_1), file)
utils.print_and_write_2("homogeneity_2", np.mean(homogeneity_2), np.std(homogeneity_2), file)
utils.print_and_write_2("silhoutte_level_1_global", np.mean(silhoutte_level_1_global), np.std(silhoutte_level_1_global), file)
utils.print_and_write_2("silhoutte_level_2_global", np.mean(silhoutte_level_2_global), np.std(silhoutte_level_2_global), file)
utils.print_and_write_2("silhoutte_level_1_weighted", np.mean(silhoutte_level_1_weighted), np.std(silhoutte_level_1_weighted), file)
utils.print_and_write_2("silhoutte_level_2_weighted", np.mean(silhoutte_level_2_weighted), np.std(silhoutte_level_2_weighted), file)
utils.print_and_write_2("dunn1_level1", np.mean(dunn1_level_1), np.std(dunn1_level_1), file)
utils.print_and_write_2("dunn2_level1", np.mean(dunn2_level_1), np.std(dunn2_level_1), file)
utils.print_and_write_2("dunn3_level1", np.mean(dunn3_level_1), np.std(dunn3_level_1), file)
utils.print_and_write_2("dunn1_level2", np.mean(dunn1_level_2), np.std(dunn1_level_2), file)
utils.print_and_write_2("dunn2_level2", np.mean(dunn2_level_2), np.std(dunn2_level_2), file)
utils.print_and_write_2("dunn3_level2", np.mean(dunn3_level_2), np.std(dunn3_level_2), file)
list_of_dtw_values = []
list_of_norm_dtw_values = []
list_of_lengths = []
data_cache = {}
for demonstration in list_of_demonstrations:
try:
list_of_frms_demonstration = list_of_frms[demonstration]
data = {}
for i in range(len(list_of_frms_demonstration)):
data[i] = [elem[0] for elem in list_of_frms_demonstration[i]]
dtw_score, normalized_dtw_score, length, labels_manual_d, colors_manual_d, labels_automatic_d, colors_automatic_d = broken_barh.plot_broken_barh(demonstration, data,
constants.PATH_TO_CLUSTERING_RESULTS + demonstration +"_" + filename + ".jpg", constants.N_COMPONENTS_TIME_ZW)
list_of_dtw_values.append(dtw_score)
list_of_norm_dtw_values.append(normalized_dtw_score)
list_of_lengths.append(length)
# Inserting manual and annotations labels into data struct before dumping as pickle file
cache_entry = {}
cache_entry['changepoints'] = list_of_frms_demonstration
cache_entry['plot_labels_manual'] = labels_manual_d
cache_entry['plot_colors_manual'] = colors_manual_d
cache_entry['plot_labels_automatic'] = labels_automatic_d
cache_entry['plot_colors_automatic'] = colors_automatic_d
data_cache[demonstration] = cache_entry
except:
print demonstration
continue
utils.print_and_write_2("dtw_score", np.mean(list_of_dtw_values), np.std(list_of_dtw_values), file)
utils.print_and_write_2("dtw_score_normalized", np.mean(list_of_norm_dtw_values), np.std(list_of_norm_dtw_values), file)
utils.print_and_write(str(list_of_lengths), file)
pickle.dump(data_cache, open(constants.PATH_TO_CLUSTERING_RESULTS + filename + "_.p", "wb"))
def post_evaluation(metrics, file, fname, vision_mode):
mutual_information_1 = []
normalized_mutual_information_1 = []
adjusted_mutual_information_1 = []
homogeneity_1 = []
precision_1_micro = []
recall_1_micro = []
precision_1_macro = []
recall_1_macro = []
precision_1_weighted = []
recall_1_weighted = []
silhouette_scores_global = []
silhouette_scores_weighted = []
dunn1_level_1 = []
dunn2_level_1 = []
dunn3_level_1 = []
list_of_frms = {}
for elem in metrics:
mutual_information_1.append(elem[0]["mutual_info_score"])
normalized_mutual_information_1.append(elem[0]["normalized_mutual_info_score"])
adjusted_mutual_information_1.append(elem[0]["adjusted_mutual_info_score"])
homogeneity_1.append(elem[0]["homogeneity_score"])
silhouette_scores_global.append(elem[1])
silhouette_scores_weighted.append(elem[6])
dunn1_level_1.append(elem[2]["level1"])
dunn2_level_1.append(elem[3]["level1"])
dunn3_level_1.append(elem[4]["level1"])
precision_1_micro.append(elem[0]["precision_micro"])
precision_1_macro.append(elem[0]["precision_macro"])
precision_1_weighted.append(elem[0]["precision_weighted"])
recall_1_micro.append(elem[0]["recall_micro"])
recall_1_macro.append(elem[0]["recall_macro"])
recall_1_weighted.append(elem[0]["recall_weighted"])
viz = elem[5]
for demonstration in viz.keys():
utils.dict_insert_list(demonstration, viz[demonstration], list_of_frms)
utils.print_and_write_2("precision_1_micro", np.mean(precision_1_micro), np.std(precision_1_micro), file)
utils.print_and_write_2("precision_1_macro", np.mean(precision_1_macro), np.std(precision_1_macro), file)
utils.print_and_write_2("precision_1_weighted", np.mean(precision_1_weighted), np.std(precision_1_weighted), file)
utils.print_and_write_2("recall_1_micro", np.mean(recall_1_micro), np.std(recall_1_micro), file)
utils.print_and_write_2("recall_1_macro", np.mean(recall_1_macro), np.std(recall_1_macro), file)
utils.print_and_write_2("recall_1_weighted", np.mean(recall_1_weighted), np.std(recall_1_weighted), file)
utils.print_and_write_2("mutual_info", np.mean(mutual_information_1), np.std(mutual_information_1), file)
utils.print_and_write_2("normalized_mutual_info", np.mean(normalized_mutual_information_1), np.std(normalized_mutual_information_1), file)
utils.print_and_write_2("adjusted_mutual_info", np.mean(adjusted_mutual_information_1), np.std(adjusted_mutual_information_1), file)
utils.print_and_write_2("silhouette_scores_global", np.mean(silhouette_scores_global), np.std(silhouette_scores_global), file)
utils.print_and_write_2("silhouette_scores_weighted", np.mean(silhouette_scores_weighted), np.std(silhouette_scores_weighted), file)
utils.print_and_write_2("homogeneity", np.mean(homogeneity_1), np.std(homogeneity_1), file)
utils.print_and_write_2("dunn1", np.mean(dunn1_level_1), np.std(dunn1_level_1), file)
utils.print_and_write_2("dunn2", np.mean(dunn2_level_1), np.std(dunn2_level_1), file)
utils.print_and_write_2("dunn3", np.mean(dunn3_level_1), np.std(dunn3_level_1), file)
list_of_dtw_values = []
list_of_norm_dtw_values = []
list_of_lengths = []
if vision_mode:
T = constants.N_COMPONENTS_TIME_Z
else:
T = constants.N_COMPONENTS_TIME_W
pickle.dump(list_of_frms, open(constants.PATH_TO_CLUSTERING_RESULTS + fname + "_.p", "wb"))
for demonstration in list_of_demonstrations:
try:
list_of_frms_demonstration = list_of_frms[demonstration]
data = {}
for i in range(len(list_of_frms_demonstration)):
data[i] = [elem[0] for elem in list_of_frms_demonstration[i]]
dtw_score, normalized_dtw_score, length = broken_barh.plot_broken_barh(demonstration, data,
constants.PATH_TO_CLUSTERING_RESULTS + demonstration +"_" + fname + ".jpg", T)
list_of_dtw_values.append(dtw_score)
list_of_norm_dtw_values.append(normalized_dtw_score)
list_of_lengths.append(length)
except:
print demonstration
pass
utils.print_and_write_2("dtw_score", np.mean(list_of_dtw_values), np.std(list_of_dtw_values), file)
utils.print_and_write_2("dtw_score_normalized", np.mean(list_of_norm_dtw_values), np.std(list_of_norm_dtw_values), file)
utils.print_and_write(str(list_of_lengths), file)
