def test__GmmAnalysis__make_gmm_models():
gmm = GmmAnalysis(configuration=o_config,
data=o_data,
names=o_config.normalized_error_names,
output_path=output_path,
max_components=max_components)
gmm.make_gmm_models()
for k in gmm.models:
assert isinstance(gmm.models[k]['obj'], GaussianMixture)
assert isinstance(k, int)
def dev__GmmAnalysis__do_aic_analysis():
gmm = GmmAnalysis(configuration=o_config,
data=o_data,
names=o_config.normalized_error_names,
output_path=output_path,
max_components=max_components)
gmm.make_gmm_models()
gmm.do_aic_analysis()
for k in gmm.models:
print(gmm.models[k])
def test__GmmAnalysis__do_cluster_analysis():
n_components = 10
gmm = GmmAnalysis(configuration=o_config,
data=o_data,
names=o_config.normalized_error_names,
output_path=output_path,
max_components=max_components)
gmm.make_gmm_models()
gmm.do_cluster_analysis(n_components=n_components)
for k in gmm.cluster_ids:
assert isinstance(k, int)
def test__GmmAnalysis__do_aic_analysis():
gmm = GmmAnalysis(configuration=o_config,
data=o_data,
names=o_config.normalized_error_names,
output_path=output_path,
max_components=max_components)
gmm.make_gmm_models()
gmm.do_aic_analysis()
for k in gmm.models:
assert isinstance(gmm.models[k]['aic'], float)
assert isinstance(gmm.aic_criteria, dict)
assert isinstance(gmm.aic_criteria['min_components'], int)
assert isinstance(gmm.aic_criteria['min_value'], float)
def dev__GmmAnalysis():
max_components = 20
n_components = 10
gmm = GmmAnalysis(configuration=o_config,
data=o_data,
names='all',
output_path=output_path,
max_components=max_components)
print(gmm.names)
gmm.make_gmm_models()
gmm.do_aic_analysis()
gmm.do_bic_analysis()
gmm.do_ic_analysis()
gmm.do_cluster_analysis(n_components=n_components)
# table__cluster_info(gmm)
# table__cluster_parameters(gmm)
# table__cluster_qois(gmm)
plot__cluster_qoi(gmm)
gmm.plot_gmm_analysis(n_components=20)
'pyposmat.kde.1.out')
o_config = PyposmatConfigurationFile()
o_config.read(filename=config_fn)
o_data = PyposmatDataFile()
o_data.read(filename=data_fn)
o_data.create_normalized_errors(
normalize_type='by_qoi_target',
qoi_targets=o_config.qoi_targets)
max_components = 10
gmm = GmmAnalysis(configuration=o_config,
data=o_data,
names=o_config.normalized_error_names)
gmm.make_gmm_models(max_components=max_components)
o_data.df['score'] = o_data.df[o_config.normalized_error_names].abs().sum(axis=1)
exit()
# do AIC and BIC analysis
if True:
name_1 = o_config.qoi_names[0]
name_2 = o_config.qoi_names[1]
data = o_data.df[[name_1,name_2]]
max_components = 21
gmm_analysis(
config_fn=config_fn,
data_fn=data_fn,
names=[name_1,name_2],
output_directory='gmm_analysis',
max_components=20)