def three_models_combined(self):
hypertension_features = [
'trait1', 'trait2', 'lot1', 'PCL1', 'PCL2', 'phq2',
'active_coping1', 'self_blame1',
'HL_MAOA', 'HML_FKBP5', 'highschool_diploma', 'T1Acc1n', 'T1Acc1t',
'q6.15_CONC_pcl1', 'q6.2_DREAM_pcl2',
'hypertention_pcl1', 'hypertention_pcl2', 'avoidance_pcl2', 'intrusion_pcl1', 'depression_pcl2'
]
avoidance_features = ['q6.1_INTRU_pcl1', 'q6.2_DREAM_pcl2', 'q6.3_FLASH_pcl2','q6.3_FLASH_pcl1',
'q6.4_UPSET_pcl1', 'q6.14_ANGER_pcl1', 'q6.7_AVSIT_pcl1', 'q6.7_AVSIT_pcl2',
'q6.11_NUMB_pcl1', 'q6.12_FUTRE_pcl2', 'q6.14_ANGER_pcl1',
'avoidance_pcl1', 'avoidance_pcl2', 'depression_pcl1', 'intrusion_pcl1',
'PCL_Broad2', 'PCL_Strict1','trait2'
]
intrusion_features = ['trait1', 'q6.5_PHYS_pcl1', 'q6.14_ANGER_pcl2', 'state1', 'PCL1', 'phq1', 'self_distraction1',
'hypertention_pcl2', 'venting1', 'PCL2', 'self_distraction2', 'behavioral_disengagement2',
'q6.17_STRTL_pcl2', 'substance_use1', 'HML_NPY', 'venting2', 'behavioral_disengagement1',
'ADHD', 'cd_risc1']
regression_features = [
'q6.11_NUMB_pcl2', 'q6.13_SLEEP_pcl1', 'intrusion_pcl2', 'phq2',
'q6.1_INTRU_pcl2', 'PCL_Broad1', 'q6.14_ANGER_pcl2',
'phq1', 'q6.5_PHYS_pcl1', 'denial2'
]
depression_features= ['q6.1_INTRU_pcl1', 'q6.2_DREAM_pcl2', 'q6.3_FLASH_pcl2', 'q6.3_FLASH_pcl1',
'q6.4_UPSET_pcl1', 'q6.14_ANGER_pcl1', 'q6.7_AVSIT_pcl1', 'q6.7_AVSIT_pcl2',
'q6.11_NUMB_pcl1', 'q6.12_FUTRE_pcl2', 'q6.14_ANGER_pcl1',
'avoidance_pcl1', 'avoidance_pcl2', 'depression_pcl1', 'intrusion_pcl1',
'PCL_Broad2', 'PCL_Strict1', 'trait2'
]
target = "PCL_Strict3"
multiple_features_eda = EDAMultiFeatureBackend(self.df, self.features, target)
multiple_features_eda.three_models_combined(intrusion_features, avoidance_features, hypertension_features,
regression_features, depression_features)
def test_algorithms_for_target_regression(self):
print("\n\n\n regression_cutoff \n")
f = ['q6.11_NUMB_pcl2', 'q6.13_SLEEP_pcl1', 'intrusion_pcl2', 'phq2',
'q6.1_INTRU_pcl2', 'PCL_Broad1', 'q6.14_ANGER_pcl2',
'phq1', 'q6.5_PHYS_pcl1', 'denial2'] # 20
target = "PCL3"
multiple_features_eda = EDAMultiFeatureBackend(self.df, f, target)
multiple_features_eda.regression_model()
def test_algorithms_for_target_regression(self):
print("\n\n\n regression_cutoff \n")
f = [
'q6.11_NUMB_pcl2', 'q6.13_SLEEP_pcl1', 'intrusion_pcl2', 'phq2',
'q6.1_INTRU_pcl2', 'PCL_Broad1', 'q6.14_ANGER_pcl2', 'phq1',
'q6.5_PHYS_pcl1', 'denial2'
] # 20
target = "PCL3"
multiple_features_eda = EDAMultiFeatureBackend(self.df, f, target)
multiple_features_eda.regression_model()
def test_models_with_LOO(self):
targets = {
'intrusion_cutoff': 1,
'avoidance_cutoff': 1,
'hypertention_cutoff': 1,
'depression_cutoff': 1,
'only_avoidance_cutoff': 1,
'diagnosis': 1,
'regression_cutoff_33': 1,
'regression_cutoff_50': 1,
'tred_cutoff': 1
}
targets_list = [i for i in targets if targets[i] == 1]
for target in targets_list:
print(f"\n\n\n\t\b{target}\n")
multiple_features_eda = EDAMultiFeatureBackend(self.df, self.features + self.features_2, target)
multiple_features_eda.model_selection_by_grid_search_loo()
def three_models_combined(self):
hypertension_features = [
'trait1', 'trait2', 'lot1', 'PCL1', 'PCL2', 'phq2',
'active_coping1', 'self_blame1', 'HL_MAOA', 'HML_FKBP5',
'highschool_diploma', 'T1Acc1n', 'T1Acc1t', 'q6.15_CONC_pcl1',
'q6.2_DREAM_pcl2', 'hypertention_pcl1', 'hypertention_pcl2',
'avoidance_pcl2', 'intrusion_pcl1', 'depression_pcl2'
]
avoidance_features = [
'q6.1_INTRU_pcl1', 'q6.2_DREAM_pcl2', 'q6.3_FLASH_pcl2',
'q6.3_FLASH_pcl1', 'q6.4_UPSET_pcl1', 'q6.14_ANGER_pcl1',
'q6.7_AVSIT_pcl1', 'q6.7_AVSIT_pcl2', 'q6.11_NUMB_pcl1',
'q6.12_FUTRE_pcl2', 'q6.14_ANGER_pcl1', 'avoidance_pcl1',
'avoidance_pcl2', 'depression_pcl1', 'intrusion_pcl1',
'PCL_Broad2', 'PCL_Strict1', 'trait2'
]
intrusion_features = [
'trait1', 'q6.5_PHYS_pcl1', 'q6.14_ANGER_pcl2', 'state1', 'PCL1',
'phq1', 'self_distraction1', 'hypertention_pcl2', 'venting1',
'PCL2', 'self_distraction2', 'behavioral_disengagement2',
'q6.17_STRTL_pcl2', 'substance_use1', 'HML_NPY', 'venting2',
'behavioral_disengagement1', 'ADHD', 'cd_risc1'
]
regression_features = [
'q6.11_NUMB_pcl2', 'q6.13_SLEEP_pcl1', 'intrusion_pcl2', 'phq2',
'q6.1_INTRU_pcl2', 'PCL_Broad1', 'q6.14_ANGER_pcl2', 'phq1',
'q6.5_PHYS_pcl1', 'denial2'
]
depression_features = [
'q6.1_INTRU_pcl1', 'q6.2_DREAM_pcl2', 'q6.3_FLASH_pcl2',
'q6.3_FLASH_pcl1', 'q6.4_UPSET_pcl1', 'q6.14_ANGER_pcl1',
'q6.7_AVSIT_pcl1', 'q6.7_AVSIT_pcl2', 'q6.11_NUMB_pcl1',
'q6.12_FUTRE_pcl2', 'q6.14_ANGER_pcl1', 'avoidance_pcl1',
'avoidance_pcl2', 'depression_pcl1', 'intrusion_pcl1',
'PCL_Broad2', 'PCL_Strict1', 'trait2'
]
target = "PCL_Strict3"
multiple_features_eda = EDAMultiFeatureBackend(self.df, self.features,
target)
multiple_features_eda.three_models_combined(intrusion_features,
avoidance_features,
hypertension_features,
regression_features,
depression_features)
def test_models_with_LOO(self):
targets = {
'intrusion_cutoff': 1,
'avoidance_cutoff': 1,
'hypertention_cutoff': 1,
'depression_cutoff': 1,
'only_avoidance_cutoff': 1,
'diagnosis': 1,
'regression_cutoff_33': 1,
'regression_cutoff_50': 1,
'tred_cutoff': 1
}
targets_list = [i for i in targets if targets[i] == 1]
for target in targets_list:
print(f"\n\n\n\t\b{target}\n")
multiple_features_eda = EDAMultiFeatureBackend(
self.df, self.features + self.features_2, target)
multiple_features_eda.model_selection_by_grid_search_loo()
def test_algorithms_for_target_tred_cutoff(self):
print("precision score")
print("\n\n\n tred_cutoff \n")
target = "tred_cutoff"
multiple_features_eda = EDAMultiFeatureBackend(
self.df, self.features + self.features_2, target)
print("\n tred_cutoff with resample\nn=10")
#multiple_features_eda.model_checking(10, scoring='precision')
print("\n tred_cutoff without resample\nn=10")
#multiple_features_eda.model_checking_without_resampling(10, scoring='precision')
print("\n tred_cutoff with resample\nn=20")
#multiple_features_eda.model_checking(20, scoring='precision')
print("\n tred_cutoff without resample\nn=20")
#multiple_features_eda.model_checking_without_resampling(20, scoring='precision')
print("\n tred_cutoff with resample\nn=30")
#multiple_features_eda.model_checking(30, scoring='precision')
print("\n tred_cutoff without resample\nn=30")
#multiple_features_eda.model_checking_without_resampling(30, scoring='precision')
print("tred_cutoff logreg f1 score")
multiple_features_eda.logistic_regression_grid_search()
print("tred_cutoff logreg precision score")
multiple_features_eda.logistic_regression_grid_search(
scoring='precision')
def test_algorithms_for_target_tred_cutoff(self):
print("precision score")
print("\n\n\n tred_cutoff \n")
target = "tred_cutoff"
multiple_features_eda = EDAMultiFeatureBackend(self.df, self.features + self.features_2, target)
print("\n tred_cutoff with resample\nn=10")
#multiple_features_eda.model_checking(10, scoring='precision')
print("\n tred_cutoff without resample\nn=10")
#multiple_features_eda.model_checking_without_resampling(10, scoring='precision')
print("\n tred_cutoff with resample\nn=20")
#multiple_features_eda.model_checking(20, scoring='precision')
print("\n tred_cutoff without resample\nn=20")
#multiple_features_eda.model_checking_without_resampling(20, scoring='precision')
print("\n tred_cutoff with resample\nn=30")
#multiple_features_eda.model_checking(30, scoring='precision')
print("\n tred_cutoff without resample\nn=30")
#multiple_features_eda.model_checking_without_resampling(30, scoring='precision')
print("tred_cutoff logreg f1 score")
multiple_features_eda.logistic_regression_grid_search()
print("tred_cutoff logreg precision score")
multiple_features_eda.logistic_regression_grid_search(scoring='precision')
def test_algorithms_for_target_diagnosis(self):
print("precision score")
print("\n\n\n diagnosis \n")
target = "diagnosis"
multiple_features_eda = EDAMultiFeatureBackend(self.df, self.features + self.features_2, target)
print("\n diagnosis with resample\nn=10")
multiple_features_eda.model_checking(10, scoring='precision')
print("\n diagnosis without resample\nn=10")
multiple_features_eda.model_checking_without_resampling(10, scoring='precision')
print("\n diagnosis with resample\nn=20")
multiple_features_eda.model_checking(20, scoring='precision')
print("\n diagnosis without resample\nn=20")
multiple_features_eda.model_checking_without_resampling(20, scoring='precision')
print("\n diagnosis with resample\nn=30")
multiple_features_eda.model_checking(30, scoring='precision')
print("\n diagnosis without resample\nn=30")
multiple_features_eda.model_checking_without_resampling(30, scoring='precision')
def test_algorithms_for_target_diagnosis(self):
print("precision score")
print("\n\n\n diagnosis \n")
target = "diagnosis"
multiple_features_eda = EDAMultiFeatureBackend(
self.df, self.features + self.features_2, target)
print("\n diagnosis with resample\nn=10")
multiple_features_eda.model_checking(10, scoring='precision')
print("\n diagnosis without resample\nn=10")
multiple_features_eda.model_checking_without_resampling(
10, scoring='precision')
print("\n diagnosis with resample\nn=20")
multiple_features_eda.model_checking(20, scoring='precision')
print("\n diagnosis without resample\nn=20")
multiple_features_eda.model_checking_without_resampling(
20, scoring='precision')
print("\n diagnosis with resample\nn=30")
multiple_features_eda.model_checking(30, scoring='precision')
print("\n diagnosis without resample\nn=30")
multiple_features_eda.model_checking_without_resampling(
30, scoring='precision')