def exct_mtf_autoencoder_hierarchy(n_clicks, mtf_data, ae_data, hrc_data):
print("MTF Autoencoder & Hierarchy")
# init
result, result_norm = initialize_data()
result_resample = exec_ts_resampler(result_norm, mtf_data[0]['image_size'])
#(242,28,1)
result_ = result_resample.reshape(result_resample.shape[0], 1,
result_resample.shape[1])
#(242,28,28)
X = toMTFdata(tsdatas=result_,
image_size=mtf_data[0]['image_size'],
n_bins=mtf_data[0]['n_bins'],
strategy=mtf_data[0]['mtf_strategy'])
X_expand = np.expand_dims(X, axis=3)
all_feature = fit_autoencoder(
X_expand, mtf_data[0]['image_size'], ae_data[0]['dimension_feature'],
ae_data[0]['optimizer'], (3e-7) * (10**ae_data[0]['learning_rate']),
ae_data[0]['activation_function'], ae_data[0]['loss_function'],
ae_data[0]['batch_size'], ae_data[0]['epoch'])
print(f'feature shape{all_feature.shape}')
cluster = hierarchicalClustering(all_feature,
hrc_data[0]['number_of_cluster'],
hrc_data[0]['linkage'])
send_result_data(result, hrc_data[0]['number_of_cluster'],
"MTF Autoencoder & Hierarchy", cluster.labels_,
all_feature)
return []
def exct_gaf_autoencoder_dbscan(n_clicks, gaf_data, ae_data, dbs_data):
print("GAF Autoencoder & DBSCAN 실행중입니다...")
# init
result, result_norm = initialize_data()
result_resample = exec_ts_resampler(result_norm, gaf_data[0]['image_size'])
#(242,28,1)
result_ = result_resample.reshape(result_resample.shape[0], 1,
result_resample.shape[1])
#(242,28,28)
X = toGAFdata(tsdatas=result_,
image_size=gaf_data[0]['image_size'],
method=gaf_data[0]['gaf_method'])
X_expand = np.expand_dims(X, axis=3)
all_feature = fit_autoencoder(
X_expand, gaf_data[0]['image_size'], ae_data[0]['dimension_feature'],
ae_data[0]['optimizer'], (3e-7) * (10**ae_data[0]['learning_rate']),
ae_data[0]['activation_function'], ae_data[0]['loss_function'],
ae_data[0]['batch_size'], ae_data[0]['epoch'])
cluster = dbscan(all_feature,
eps=dbs_data[0]['epsilon'],
min_samples=dbs_data[0]['min_sample'])
cluster_num = max(cluster.labels_) + 1
send_result_data(result, cluster_num, "GAF Autoencoder & DBSCAN",
cluster.labels_, all_feature)
# init
result, result_norm = initialize_data()
return []
def exct_rp_autoencoder_dbscan(n_clicks, rp_data, ae_data, dbs_data):
print("RP Autoencoder & DBSCAN 실행중 입니다...")
threshold = rp_data[0]['threshold']
if threshold == 'None':
threshold = None
# init
result, result_norm = initialize_data()
result_resample = exec_ts_resampler(result_norm, rp_data[0]['image_size'])
#(242,28,1)
result_ = result_resample.reshape(result_resample.shape[0], 1,
result_resample.shape[1])
#(242,28,28)
X = toRPdata(result_, rp_data[0]['dimension'], rp_data[0]['time_delay'],
threshold, rp_data[0]['percentage'] / 100)
X_expand = np.expand_dims(X, axis=3)
all_feature = fit_autoencoder(
X_expand, rp_data[0]['image_size'], ae_data[0]['dimension_feature'],
ae_data[0]['optimizer'], (3e-7) * (10**ae_data[0]['learning_rate']),
ae_data[0]['activation_function'], ae_data[0]['loss_function'],
ae_data[0]['batch_size'], ae_data[0]['epoch'])
cluster = dbscan(all_feature,
eps=dbs_data[0]['epsilon'],
min_samples=dbs_data[0]['min_sample'])
cluster_num = max(cluster.labels_) + 1
send_result_data(result, cluster_num, "RP Autoencoder & DBSCAN",
cluster.labels_, all_feature)
return []
def exct_gaf_autoencoder_kmeans(n_clicks, gaf_data, ae_data, km_data):
print("GAF Autoencoder & Kmeans 실행중입니다...")
# init
result, result_norm = initialize_data()
result_resample = exec_ts_resampler(result_norm, gaf_data[0]['image_size'])
#(242,28,1)
result_ = result_resample.reshape(result_resample.shape[0], 1,
result_resample.shape[1])
#(242,28,28)
X = toGAFdata(tsdatas=result_,
image_size=gaf_data[0]['image_size'],
method=gaf_data[0]['gaf_method'])
X_expand = np.expand_dims(X, axis=3)
all_feature = fit_autoencoder(
X_expand, gaf_data[0]['image_size'], ae_data[0]['dimension_feature'],
ae_data[0]['optimizer'], (3e-7) * (10**ae_data[0]['learning_rate']),
ae_data[0]['activation_function'], ae_data[0]['loss_function'],
ae_data[0]['batch_size'], ae_data[0]['epoch'])
print(f'feature shape{all_feature.shape}')
cluster = kmeans(all_feature, km_data[0]['number_of_cluster'],
km_data[0]['tolerance'], km_data[0]['try_n_init'],
km_data[0]['try_n_kmeans'])
send_result_data(result, km_data[0]['number_of_cluster'],
"GAF Autoencoder & Kmeans", cluster.labels_, all_feature)
return []
def exct_rp_autoencoder_hierarchy(n_clicks, rp_data, ae_data, hrc_data):
print("RP Autoencoder & hierarchy 실행중입니다...")
threshold = rp_data[0]['threshold']
if threshold == 'None':
threshold = None
# init
result, result_norm = initialize_data()
result_resample = exec_ts_resampler(result_norm, rp_data[0]['image_size'])
#(242,28,1)
result_ = result_resample.reshape(result_resample.shape[0], 1,
result_resample.shape[1])
#(242,28,28)
X = toRPdata(result_, rp_data[0]['dimension'], rp_data[0]['time_delay'],
threshold, rp_data[0]['percentage'] / 100)
X_expand = np.expand_dims(X, axis=3)
all_feature = fit_autoencoder(
X_expand, rp_data[0]['image_size'], ae_data[0]['dimension_feature'],
ae_data[0]['optimizer'], (3e-7) * (10**ae_data[0]['learning_rate']),
ae_data[0]['activation_function'], ae_data[0]['loss_function'],
ae_data[0]['batch_size'], ae_data[0]['epoch'])
cluster = hierarchicalClustering(all_feature,
hrc_data[0]['number_of_cluster'],
hrc_data[0]['linkage'])
send_result_data(result, hrc_data[0]['number_of_cluster'],
"RP Autoencoder & Hierarchy", cluster.labels_,
all_feature)
return 0