def main(on_switch=False):
if on_switch:
obj = DataPreparationNameEthnicityProject()
obj.data_preparation_steps()
if obj.control_panel['done_switch']:
hf.done_alert()
def main(on_switch=False):
if on_switch:
obj = FeatureCreationNameEthnicityProject()
obj.feature_generation_steps()
if local_control_panel['done_switch']:
hf.done_alert()
def main(on_switch=False):
if on_switch:
save_switch = False # WARNING: Will overwrite existing if True
run_on_subsampled_data = True
run_on_full_data = False
run_on_unfeatured_data = False
run_on_featured_data = True
# Eg, full list >> [1, 5, 10, 50, 100, 500]
nk_list = [5]
'''Eg, Full list >> ['dummy', 'sex_only', 'name_basic_only', 'name_substring_only', 'name_numeric_only',
'name_metaphone_only', 'name_all', 'loc_basic_only', 'loc_sep_entity_only',
'loc_substring_only', 'loc_all', 'name_all_loc_all']'''
feature_set_list = ['name_all_loc_all_reduced']
ml_algo_dict = {
'LR': {
'clf':
LogisticRegression(tol=3.359818286283781e-05,
solver='liblinear',
penalty='l1',
max_iter=4000,
class_weight=None,
C=4.281332398719396)
},
'SVC_LINEAR': {
'clf': CalibratedClassifierCV(LinearSVC(penalty='l2', C=0.01)),
},
'NB': {
'clf': BernoulliNB(),
},
}
# Eg, Full list >> ['ab', 'fn', 'metis', 'inuit', 'ch', 'ja', 'en', 'fr', 'ir', 'it', 'rus', 'sc', 'others']
target_set_list = ['ch', 'ab']
if run_on_subsampled_data:
# Loop through subsampling n set with unfeatured data
if run_on_unfeatured_data:
for nk in nk_list:
for algo_key, algo_val in ml_algo_dict.items():
for target_label in target_set_list:
print('>> Current time:', datetime.datetime.now())
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(
control_panel={
'save_result_switch':
save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch':
False, # WARNING: Switch to False in production
'use_subsampled_df_nk': nk,
'use_featured_df_switch': False,
'use_feature_set': [],
'feature_selection_switch': False,
'cross_validation_switch': False,
'ml_process_on_test_data_switch': True,
'ml_process_on_training_data_switch':
False,
'ml_process_on_ext_data_switch': False,
'ml_algo': [algo_key, algo_val],
'ml_algo_param_grid': None,
'binary_target_label': target_label,
'eval_score': None,
'random_state': 888,
})
obj.machine_learning_steps()
# Loop through feature set and subsampling n set
if run_on_featured_data:
for feature_set in feature_set_list:
for nk in nk_list:
for algo_key, algo_val in ml_algo_dict.items():
for target_label in target_set_list:
print('>> Current time:',
datetime.datetime.now())
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(
control_panel={
'save_result_switch':
save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch':
False, # WARNING: Switch to False in production
'use_subsampled_df_nk': nk,
'use_featured_df_switch': True,
'use_feature_set': feature_set,
'feature_selection_switch': False,
'cross_validation_switch': False,
'ml_process_on_test_data_switch': True,
'ml_process_on_training_data_switch':
False,
'ml_process_on_ext_data_switch': False,
'ml_algo': [algo_key, algo_val],
'ml_algo_param_grid': None,
'binary_target_label': target_label,
'eval_score': None,
'random_state': 888,
})
obj.machine_learning_steps()
if run_on_full_data:
# Run once using unfeatured, full dataset
if run_on_unfeatured_data:
for feature_set in feature_set_list:
for algo_key, algo_val in ml_algo_dict.items():
for target_label in target_set_list:
print('>> Current time:', datetime.datetime.now())
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(
control_panel={
'save_result_switch':
save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch':
False, # WARNING: Switch to False in production
'use_subsampled_df_nk': 'none',
'use_featured_df_switch': False,
'use_feature_set': [],
'feature_selection_switch': False,
'cross_validation_switch': False,
'ml_process_on_test_data_switch': True,
'ml_process_on_training_data_switch':
False,
'ml_process_on_ext_data_switch': False,
'ml_algo': [algo_key, algo_val],
'ml_algo_param_grid': None,
'binary_target_label': target_label,
'eval_score': None,
'random_state': 888,
})
obj.machine_learning_steps()
# Run once using featured, full dataset
if run_on_featured_data:
for feature_set in feature_set_list:
for algo_key, algo_val in ml_algo_dict.items():
for target_label in target_set_list:
print('>> Current time:', datetime.datetime.now())
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(
control_panel={
'save_result_switch':
save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch':
False, # WARNING: Switch to False in production
'use_subsampled_df_nk': [],
'use_featured_df_switch': True,
'use_feature_set': feature_set,
'feature_selection_switch': False,
'cross_validation_switch': False,
'ml_process_on_test_data_switch': True,
'ml_process_on_training_data_switch':
False,
'ml_process_on_ext_data_switch': False,
'ml_algo': [algo_key, algo_val],
'ml_algo_param_grid': None,
'binary_target_label': target_label,
'eval_score': None,
'random_state': 888,
})
obj.machine_learning_steps()
if local_control_panel['done_switch']:
hf.done_alert()
def main(on_switch=False):
if on_switch:
save_switch = False
run_on_subsampled_data = False
run_on_full_data = True
run_on_unfeatured_data = False
run_on_featured_data = True
# Eg, full list >> [1, 5, 10, 50, 100, 500]
nk_list = [500]
'''Eg, Full list >> ['dummy', 'sex_only', 'name_basic_only', 'name_substring_only', 'name_numeric_only',
'name_metaphone_only', 'name_all', 'loc_basic_only', 'loc_sep_entity_only',
'loc_substring_only', 'loc_all', 'name_all_loc_all']'''
feature_set_list = ['dummy', 'name_all', 'name_all_loc_all_reduced']
eval_score_list = ['macro f1 score']
target_label_list = ['ab', 'fn', 'metis', 'inuit', 'ch', 'ja', 'fr']
ml_algo_param_dict = \
{ 'LR': { 'clf': LogisticRegression(),
'param': { # empty since we're not doing grid-search here
}},
}
if run_on_subsampled_data:
# Loop through subsampling n set with unfeatured data
if run_on_unfeatured_data:
for nk in nk_list:
for algo_key, algo_val in ml_algo_param_dict.items():
for target_label in target_label_list:
for eval_score in eval_score_list:
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False, # WARNING: Switch to False in production
'use_subsampled_df_nk': nk,
'use_featured_df_switch': False,
'use_feature_set': 'none',
'feature_selection_switch': True,
'cross_validation_switch': False,
'ml_process_on_test_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'binary_target_label': target_label,
'eval_score': eval_score,
'random_state': 888,
})
obj.machine_learning_steps()
# Loop through feature set and subsampling n set
if run_on_featured_data:
for feature_set in feature_set_list:
for nk in nk_list:
for algo_key, algo_val in ml_algo_param_dict.items():
for target_label in target_label_list:
for eval_score in eval_score_list:
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False, # WARNING: Switch to False in production
'use_subsampled_df_nk': nk,
'use_featured_df_switch': True,
'use_feature_set': feature_set,
'feature_selection_switch': True,
'cross_validation_switch': False,
'ml_process_on_test_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'binary_target_label': target_label,
'eval_score': eval_score,
'random_state': 888,
})
obj.machine_learning_steps()
if run_on_full_data:
# Run once using unfeatured, full dataset
if run_on_unfeatured_data:
for feature_set in feature_set_list:
for algo_key, algo_val in ml_algo_param_dict.items():
for target_label in target_label_list:
for eval_score in eval_score_list:
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False, # WARNING: Switch to False in production
'use_subsampled_df_nk': 'none',
'use_featured_df_switch': False,
'use_feature_set': 'none',
'feature_selection_switch': True,
'cross_validation_switch': False,
'ml_process_on_test_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'binary_target_label': target_label,
'eval_score': eval_score,
'random_state': 888,
})
obj.machine_learning_steps()
# Run once using featured, full dataset
if run_on_featured_data:
for feature_set in feature_set_list:
for algo_key, algo_val in ml_algo_param_dict.items():
for target_label in target_label_list:
for eval_score in eval_score_list:
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False, # WARNING: Switch to False in production
'use_subsampled_df_nk': 'none',
'use_featured_df_switch': True,
'use_feature_set': feature_set,
'feature_selection_switch': True,
'cross_validation_switch': False,
'ml_process_on_test_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'binary_target_label': target_label,
'eval_score': eval_score,
'random_state': 888,
})
obj.machine_learning_steps()
if local_control_panel['done_switch']:
hf.done_alert()
def main(on_switch=False):
if on_switch:
save_switch = False
run_on_subsampled_data = True
run_on_full_data = False
run_on_unfeatured_data = False
run_on_featured_data = True
# Eg, full list >> [1, 5, 10, 50, 100, 500]
nk_list = [5]
cv_repeat = 1
'''Eg, Full list >> ['dummy', 'sex_only', 'name_basic_only', 'name_substring_only', 'name_numeric_only',
'name_metaphone_only', 'name_all', 'loc_basic_only', 'loc_sep_entity_only',
'loc_substring_only', 'loc_all', 'name_all_loc_all', 'name_all_loc_all_reduced']'''
feature_set_list = ['name_all']
eval_score_list = ['macro f1 score']
# Eg, Full list >> ['ab', 'fn', 'metis', 'inuit', 'ch', 'ja', 'en', 'fr', 'ir', 'it', 'rus', 'sc', 'others']
target_label_list = ['fn']
ml_algo_param_dict = \
{
'LR_V1': { 'clf': LogisticRegression(),
'param': {
'logisticregression__solver': ['liblinear'],
'logisticregression__penalty': ['l1', 'l2'],
'logisticregression__C': np.logspace(-4, 4, 20),
'logisticregression__tol': np.logspace(-5, 5, 20),
'logisticregression__class_weight': [None, 'balanced'],
'logisticregression__max_iter': [50, 1000, 4000, 20000],
}},
'LR_V2': { 'clf': LogisticRegression(),
'param': {
'logisticregression__solver': ['newton-cg', 'lbfgs', 'sag', 'saga'],
'logisticregression__penalty': ['none', 'l2'],
'logisticregression__C': np.logspace(-4, 4, 20),
'logisticregression__tol': np.logspace(-5, 5, 20),
'logisticregression__class_weight': [None, 'balanced'],
'logisticregression__max_iter': [50, 1000, 4000, 20000],
}},
'SVC_LINEAR': { 'clf': LinearSVC(),
'param': {
'linearsvc__penalty': ['l2'],
'linearsvc__loss': ['hinge', 'squared_hinge'],
'linearsvc__C': np.logspace(-4, 4, 20),
'linearsvc__tol': [0.00001, 0.0001, 0.001, 0.01, 0.1, 1],
'linearsvc__class_weight': [None, 'balanced'],
'linearsvc__max_iter': [50, 1000, 4000, 20000],
}},
'SVC_NONLINEAR': { 'clf': SVC(),
'param': {
'svc__kernel': ['poly', 'rbf', 'sigmoid'],
'svc__C': np.logspace(-4, 4, 20),
'svc__tol': [0.00001, 0.0001, 0.001, 0.01, 0.1, 1],
'svc__class_weight': [None, 'balanced'],
'svc__decision_function_shape': ['ovo', 'ovr'],
'svc__max_iter': [50, 1000, 4000, 20000],
}},
'NB': { 'clf': BernoulliNB(),
'param': {
'bernoullinb__alpha': np.logspace(-4, 4, 20),
'bernoullinb__binarize': [None, 0, .2, .4, .6, .8, 1],
'bernoullinb__fit_prior': [True, False],
}},
}
if run_on_subsampled_data:
# Loop through subsampling n set with unfeatured data
if run_on_unfeatured_data:
for nk in nk_list:
for target_label in target_label_list:
for algo_key, algo_val in ml_algo_param_dict.items():
for eval_score in eval_score_list:
for i in range(1, cv_repeat+1):
print('>> Current time:', datetime.datetime.now())
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False, # WARNING: Switch to False in production
'use_subsampled_df_nk': nk,
'use_featured_df_switch': False,
'use_feature_set': [],
'feature_selection_switch': False,
'cross_validation_switch': True,
'cross_validation_repeat': i,
'ml_process_on_test_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'binary_target_label': target_label,
'eval_score': eval_score,
'random_state': 888,
})
obj.machine_learning_steps()
# Loop through feature set and subsampling n set
if run_on_featured_data:
for feature_set in feature_set_list:
for nk in nk_list:
for target_label in target_label_list:
for algo_key, algo_val in ml_algo_param_dict.items():
for eval_score in eval_score_list:
for i in range(1, cv_repeat+1):
print('>> Current time:', datetime.datetime.now())
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False, # WARNING: Switch to False in production
'use_subsampled_df_nk': nk,
'use_featured_df_switch': True,
'use_feature_set': feature_set,
'feature_selection_switch': False,
'cross_validation_switch': True,
'cross_validation_repeat': i,
'ml_process_on_test_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'binary_target_label': target_label,
'eval_score': eval_score,
'random_state': 888,
})
obj.machine_learning_steps()
if run_on_full_data:
# Run once using unfeatured, full dataset
if run_on_unfeatured_data:
for feature_set in feature_set_list:
for target_label in target_label_list:
for algo_key, algo_val in ml_algo_param_dict.items():
for eval_score in eval_score_list:
for i in range(1, cv_repeat+1):
print('>> Current time:', datetime.datetime.now())
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False, # WARNING: Switch to False in production
'use_subsampled_df_nk': 'none',
'use_featured_df_switch': False,
'use_feature_set': [],
'feature_selection_switch': False,
'cross_validation_switch': True,
'cross_validation_repeat': i,
'ml_process_on_test_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'binary_target_label': target_label,
'eval_score': eval_score,
'random_state': 888,
})
obj.machine_learning_steps()
# Run once using featured, full dataset
if run_on_featured_data:
for feature_set in feature_set_list:
for target_label in target_label_list:
for algo_key, algo_val in ml_algo_param_dict.items():
for eval_score in eval_score_list:
for i in range(1, cv_repeat+1):
print('>> Current time:', datetime.datetime.now())
obj = ml_pipeline.MachineLearningNameEthnicityProjectBinaryClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False, # WARNING: Switch to False in production
'use_subsampled_df_nk': [],
'use_featured_df_switch': True,
'use_feature_set': feature_set,
'feature_selection_switch': False,
'cross_validation_switch': True,
'cross_validation_repeat': i,
'ml_process_on_test_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'binary_target_label': target_label,
'eval_score': eval_score,
'random_state': 888,
})
obj.machine_learning_steps()
if local_control_panel['done_switch']:
hf.done_alert()
def main(on_switch=False):
if on_switch:
save_switch = False
run_on_subsampled_data = False
run_on_full_data = True
run_on_unfeatured_data = False
run_on_featured_data = True
# Eg, full list >> [5, 50, 500]
nk_list = [5]
# Eg, ['dummy', 'sex_only', 'name_all', 'name_all_loc_all', 'name_all_loc_all_reduced']
feature_set_list = ['name_all_loc_all_reduced']
ml_algo_dict = {
'LR': {'clf': LogisticRegression(
tol=0.01438449888287663,
solver='liblinear',
penalty='l2',
multi_class='ovr',
max_iter=50,
class_weight=None,
C=10000.0)
},
'SVC': {'clf': OneVsRestClassifier(LinearSVC(
tol=0.0001,
penalty='l2',
multi_class='ovr',
max_iter=50,
loss='squared_hinge',
class_weight='balanced',
C=0.08858667904100823)),
},
'NB': {'clf': BernoulliNB(
fit_prior=True,
binarize=0.6,
alpha=0.00026366508987303583)
},
}
if run_on_subsampled_data:
# Loop through subsampling n set with unfeatured data
if run_on_unfeatured_data:
for nk in nk_list:
for algo_key, algo_val in ml_algo_dict.items():
obj = ml_pipeline.MachineLearningNameEthnicityProjectMultiClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': True,
'use_subsampled_df_nk': nk,
'use_featured_df_switch': False,
'use_feature_set': [],
'feature_selection_switch': False,
'cross_validation_switch': False,
'cross_validation_repeat': None,
'ml_process_on_test_data_switch': True,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': [algo_key, algo_val],
'ml_algo_param_grid': None,
'eval_score': None,
'label_varname': 'ETHNICITY_RECAT',
'random_state': 888,
})
obj.machine_learning_steps()
# Loop through feature set and subsampling n set
if run_on_featured_data:
for feature_set in feature_set_list:
for nk in nk_list:
for algo_key, algo_val in ml_algo_dict.items():
obj = ml_pipeline.MachineLearningNameEthnicityProjectMultiClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': True,
'use_subsampled_df_nk': nk,
'use_featured_df_switch': True,
'use_feature_set': feature_set,
'feature_selection_switch': False,
'cross_validation_switch': False,
'cross_validation_repeat': None,
'ml_process_on_test_data_switch': True,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': [algo_key, algo_val],
'ml_algo_param_grid': None,
'eval_score': None,
'label_varname': 'ETHNICITY_RECAT',
'random_state': 888,
})
obj.machine_learning_steps()
if run_on_full_data:
# Run once using unfeatured, full dataset
if run_on_unfeatured_data:
for algo_key, algo_val in ml_algo_dict.items():
obj = ml_pipeline.MachineLearningNameEthnicityProjectMultiClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False,
'use_subsampled_df_nk': [],
'use_featured_df_switch': True,
'use_feature_set': [],
'feature_selection_switch': False,
'cross_validation_switch': False,
'cross_validation_repeat': None,
'ml_process_on_test_data_switch': True,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': [algo_key, algo_val],
'ml_algo_param_grid': None,
'eval_score': None,
'label_varname': 'ETHNICITY_RECAT',
'random_state': 888,
})
obj.machine_learning_steps()
# Run once using featured, full dataset
if run_on_featured_data:
for feature_set in feature_set_list:
for algo_key, algo_val in ml_algo_dict.items():
obj = ml_pipeline.MachineLearningNameEthnicityProjectMultiClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False,
'use_subsampled_df_nk': [],
'use_featured_df_switch': True,
'use_feature_set': feature_set,
'feature_selection_switch': False,
'cross_validation_switch': False,
'cross_validation_repeat': None,
'ml_process_on_test_data_switch': True,
'ml_process_on_training_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_algo': [algo_key, algo_val],
'ml_algo_param_grid': None,
'eval_score': None,
'label_varname': 'ETHNICITY_RECAT',
'random_state': 888,
})
obj.machine_learning_steps()
if local_control_panel['done_switch']:
hf.done_alert()
def main(on_switch=False):
if on_switch:
save_switch = False
run_on_subsampled_data = False
run_on_full_data = True
run_on_unfeatured_data = False
run_on_featured_data = True
# Eg, full list >> [5, 50, 500]
nk_list = [5]
cv_repeat = 1
# Eg = ['name_all', 'name_all_loc_all', 'name_all_loc_all_reduced']
feature_set_list = ['name_all_loc_all_reduced']
# Eg = ['accuracy', 'macro f1 score', 'macro precision', 'macro recall']
eval_score_list = ['macro f1 score']
ml_algo_param_dict = \
{
'LR_V1': { 'clf': LogisticRegression(),
'param': {
'logisticregression__solver': ['liblinear'],
'logisticregression__penalty': ['l1', 'l2'],
'logisticregression__C': np.logspace(-4, 4, 20),
'logisticregression__tol': np.logspace(-5, 5, 20),
'logisticregression__class_weight': [None, 'balanced'],
'logisticregression__multi_class': ['ovr', 'auto'],
'logisticregression__max_iter': [50, 1000, 4000, 20000],
}},
'LR_V2': { 'clf': LogisticRegression(),
'param': {
'logisticregression__solver': ['newton-cg', 'lbfgs', 'sag', 'saga'],
'logisticregression__penalty': ['none', 'l2'],
'logisticregression__C': np.logspace(-4, 4, 20),
'logisticregression__tol': np.logspace(-5, 5, 20),
'logisticregression__class_weight': [None, 'balanced'],
'logisticregression__multi_class': ['ovr', 'multinomial', 'auto'],
'logisticregression__max_iter': [50, 1000, 4000, 20000],
}},
'SVC_LINEAR': { 'clf': OneVsRestClassifier(LinearSVC()),
'param': {
'onevsrestclassifier__estimator__penalty': ['l2'],
'onevsrestclassifier__estimator__loss': ['hinge', 'squared_hinge'],
'onevsrestclassifier__estimator__C': np.logspace(-4, 4, 20),
'onevsrestclassifier__estimator__tol': [0.00001, 0.0001, 0.001, 0.01, 0.1, 1],
'onevsrestclassifier__estimator__class_weight': [None, 'balanced'],
'onevsrestclassifier__estimator__multi_class': ['ovr', 'crammer_singer'],
'onevsrestclassifier__estimator__max_iter': [50, 1000, 4000, 20000],
}},
'SVC_NONLINEAR': { 'clf': OneVsRestClassifier(SVC()),
'param': {
'onevsrestclassifier__estimator__kernel': ['poly', 'rbf', 'sigmoid'],
'onevsrestclassifier__estimator__C': np.logspace(-4, 4, 20),
'onevsrestclassifier__estimator__tol': [0.00001, 0.0001, 0.001, 0.01, 0.1, 1],
'onevsrestclassifier__estimator__class_weight': [None, 'balanced'],
'onevsrestclassifier__estimator__decision_function_shape': ['ovo', 'ovr'],
'onevsrestclassifier__estimator__max_iter': [50, 1000, 4000, 20000],
}},
'NB': { 'clf': BernoulliNB(),
'param': {
'bernoullinb__alpha': np.logspace(-4, 4, 20),
'bernoullinb__binarize': [None, 0, .2, .4, .6, .8, 1],
'bernoullinb__fit_prior': [True, False],
}},
}
if run_on_subsampled_data:
# Loop through subsampling n set with unfeatured data
if run_on_unfeatured_data:
for nk in nk_list:
for algo_key, algo_val in ml_algo_param_dict.items():
for eval_score in eval_score_list:
for i in range(1, cv_repeat+1):
obj = ml_pipeline.MachineLearningNameEthnicityProjectMultiClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': True,
'use_subsampled_df_nk': nk,
'use_featured_df_switch': False,
'use_feature_set': [],
'feature_selection_switch': False,
'cross_validation_switch': True,
'cross_validation_repeat': i,
'ml_process_on_test_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'eval_score': eval_score,
'label_varname': 'ETHNICITY_RECAT',
'random_state': 888,
})
obj.machine_learning_steps()
# Loop through feature set and subsampling n set
if run_on_featured_data:
for feature_set in feature_set_list:
for nk in nk_list:
for algo_key, algo_val in ml_algo_param_dict.items():
for eval_score in eval_score_list:
for i in range(1, cv_repeat+1):
obj = ml_pipeline.MachineLearningNameEthnicityProjectMultiClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': True,
'use_subsampled_df_nk': nk,
'use_featured_df_switch': True,
'use_feature_set': feature_set,
'feature_selection_switch': False,
'cross_validation_switch': True,
'cross_validation_repeat': i,
'ml_process_on_test_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'eval_score': eval_score,
'label_varname': 'ETHNICITY_RECAT',
'random_state': 888,
})
obj.machine_learning_steps()
if run_on_full_data:
# Run once using unfeatured, full dataset
if run_on_unfeatured_data:
for algo_key, algo_val in ml_algo_param_dict.items():
for eval_score in eval_score_list:
for i in range(1, cv_repeat+1):
obj = ml_pipeline.MachineLearningNameEthnicityProjectMultiClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False,
'use_subsampled_df_nk': [],
'use_featured_df_switch': True,
'use_feature_set': [],
'feature_selection_switch': False,
'cross_validation_switch': True,
'cross_validation_repeat': i,
'ml_process_on_test_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'eval_score': eval_score,
'label_varname': 'ETHNICITY_RECAT',
'random_state': 888,
})
obj.machine_learning_steps()
# Run once using featured, full dataset
if run_on_featured_data:
for feature_set in feature_set_list:
for algo_key, algo_val in ml_algo_param_dict.items():
for eval_score in eval_score_list:
for i in range(1, cv_repeat+1):
obj = ml_pipeline.MachineLearningNameEthnicityProjectMultiClass(control_panel = {
'save_result_switch': save_switch, # WARNING: Will overwrite existing
'use_subsampled_df_switch': False,
'use_subsampled_df_nk': [],
'use_featured_df_switch': True,
'use_feature_set': feature_set,
'feature_selection_switch': False,
'cross_validation_switch': True,
'cross_validation_repeat': i,
'ml_process_on_test_data_switch': False,
'ml_process_on_ext_data_switch': False,
'ml_process_on_training_data_switch': False,
'ml_algo': None,
'ml_algo_param_grid': [algo_key, algo_val],
'eval_score': eval_score,
'label_varname': 'ETHNICITY_RECAT',
'random_state': 888,
})
obj.machine_learning_steps()
if local_control_panel['done_switch']:
hf.done_alert()
