-
Notifications
You must be signed in to change notification settings - Fork 0
/
run_rf.py
225 lines (180 loc) · 8.79 KB
/
run_rf.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
import pandas as pd
import numpy as np
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import roc_auc_score, roc_curve
import matplotlib.pyplot as plt
import seaborn as sns
import warnings
import os
from typing import Dict, Tuple, Any
from imblearn.over_sampling import SMOTE, RandomOverSampler, SVMSMOTE
from imblearn.under_sampling import RandomUnderSampler
from joblib import Parallel, delayed
from itertools import product
warnings.filterwarnings("ignore")
def get_fairness_metrics(y_true: np.ndarray, y_pred: np.ndarray, groups: pd.Series, fixed_fpr: float = 0.05) -> Tuple[float, pd.DataFrame]:
from aequitas.group import Group
aequitas_df = pd.DataFrame({
"score": y_pred,
"label_value": y_true,
"group": groups
})
g = Group()
disparities_df = g.get_crosstabs(aequitas_df, score_thresholds={"score_val": [fixed_fpr]})[0]
predictive_equality = disparities_df["fpr"].min() / disparities_df["fpr"].max()
return predictive_equality, disparities_df
def evaluate(predictions: np.ndarray, ground_truth: np.ndarray, X: pd.DataFrame, fixed_fpr: float = 0.05) -> list[float]:
fprs, tprs, thresholds = roc_curve(ground_truth, predictions)
tpr = tprs[fprs < fixed_fpr][-1]
fpr = fprs[fprs < fixed_fpr][-1]
threshold = thresholds[fprs < fixed_fpr][-1]
sorted_ages = np.sort(X["customer_age"])
young_threshold = sorted_ages[int(0.95 * len(sorted_ages))]
groups = (X["customer_age"] > young_threshold).map({True: ">young_threshold", False: "<=young_threshold"})
predictive_equality, _ = get_fairness_metrics(ground_truth, predictions, groups, fixed_fpr)
return [round(tpr, 10), round(predictive_equality, 6)]
def train_and_evaluate_random_forest(X: pd.DataFrame, y: pd.Series, X_val: pd.DataFrame, y_val: pd.Series,
X_test: pd.DataFrame, y_test: pd.Series, best_params: Dict[str, Any]) -> Dict[str, Any]:
rf_model = RandomForestClassifier(**best_params, random_state=42, n_jobs=-1)
rf_model.fit(X, y)
y_val_pred = rf_model.predict_proba(X_val)[:, 1]
val_results = evaluate(y_val_pred, y_val, X=X_val)
val_auc = roc_auc_score(y_val, y_val_pred)
y_test_pred = rf_model.predict_proba(X_test)[:, 1]
test_results = evaluate(y_test_pred, y_test, X=X_test)
test_auc = roc_auc_score(y_test, y_test_pred)
return {
'val_results': val_results,
'val_auc': val_auc,
'test_results': test_results,
'test_auc': test_auc
}
def read_data() -> Dict[str, pd.DataFrame]:
data = {
'X_train': pd.read_csv("data/X_train_oh_1.csv"),
'X_val': pd.read_csv("data/X_val_oh_1.csv"),
'X_test': pd.read_csv("data/X_test_oh_1.csv"),
'y_train': pd.read_csv("data/y_train_1.csv").iloc[:, 0],
'y_val': pd.read_csv("data/y_val_1.csv").iloc[:, 0],
'y_test': pd.read_csv("data/y_test_1.csv").iloc[:, 0],
}
return data
def apply_sampling_technique(X: pd.DataFrame, y: pd.Series, technique: str) -> Tuple[pd.DataFrame, pd.Series]:
if technique == 'SMOTE':
sampler = SMOTE(random_state=42)
elif technique == 'RandomOverSampler':
sampler = RandomOverSampler(random_state=42)
elif technique == 'RandomUnderSampler':
sampler = RandomUnderSampler(random_state=42)
elif technique == 'SVMSMOTE':
sampler = SVMSMOTE(random_state=42)
else:
return X, y
X_resampled, y_resampled = sampler.fit_resample(X, y)
return X_resampled, y_resampled
def evaluate_hyperparameters(params, X_train, y_train, X_val, y_val):
rf = RandomForestClassifier(**params, random_state=42, n_jobs=-1)
rf.fit(X_train, y_train)
y_val_pred = rf.predict_proba(X_val)[:, 1]
score = roc_auc_score(y_val, y_val_pred)
return score, params
def hyperparameter_tuning(X_train: pd.DataFrame, y_train: pd.Series, X_val: pd.DataFrame, y_val: pd.Series) -> Dict[str, Any]:
param_grid = {
'n_estimators': [600],
'max_depth': [8, 10, 12, 14],
'min_samples_split': [2, 4],
'max_features': ['sqrt', 'log2'],
'class_weight': ['balanced'],
}
param_combinations = list(product(*param_grid.values()))
param_dicts = [dict(zip(param_grid.keys(), params)) for params in param_combinations]
results = Parallel(n_jobs=-1)(
delayed(evaluate_hyperparameters)(params, X_train, y_train, X_val, y_val) for params in param_dicts
)
best_score, best_params = max(results, key=lambda x: x[0])
return best_params, best_score
def main():
os.makedirs('model_performance', exist_ok=True)
data = read_data()
X, y = data['X_train'], data['y_train']
X_val, y_val = data['X_val'], data['y_val']
sampling_techniques = ['None'] # , 'SMOTE', 'RandomOverSampler', 'RandomUnderSampler', 'SVMSMOTE'
for technique in sampling_techniques:
print(f"\nApplying {technique} sampling technique:")
if technique != 'None':
X_sampled, y_sampled = apply_sampling_technique(X, y, technique)
else:
X_sampled, y_sampled = X, y
best_params, best_val_auc = hyperparameter_tuning(X_sampled, y_sampled, X_val, y_val)
print("Best parameters found: ", best_params)
print(f"Best validation AUC: {best_val_auc:.4f}")
evaluation_results = train_and_evaluate_random_forest(
X_sampled, y_sampled, data['X_val'], data['y_val'],
data['X_test'], data['y_test'], best_params
)
print("Evaluation Results:")
print(f"Validation AUC: {evaluation_results['val_auc']:.4f}")
print(f"Validation Evaluation: TPR at 5% FPR = {evaluation_results['val_results'][0]:.4f}, Predictive Equality = {evaluation_results['val_results'][1]:.4f}")
print(f"Test AUC: {evaluation_results['test_auc']:.4f}")
print(f"Test Evaluation: TPR at 5% FPR = {evaluation_results['test_results'][0]:.4f}, Predictive Equality = {evaluation_results['test_results'][1]:.4f}")
if __name__ == "__main__":
main()
'''
Best parameters found: {'n_estimators': 600, 'max_depth': 12, 'min_samples_split': 4, 'max_features': 'log2', 'class_weight': None}
Best validation AUC: 0.8666
Evaluation Results:
Validation AUC: 0.8666
Validation Evaluation: TPR at 5% FPR = 0.4591, Predictive Equality = 0.2940
Test AUC: 0.8733
Test Evaluation: TPR at 5% FPR = 0.4850, Predictive Equality = 0.2489
Applying SMOTE sampling technique:
Best parameters found: {'n_estimators': 600, 'max_depth': 8, 'min_samples_split': 4, 'max_features': 'log2', 'class_weight': None}
Best validation AUC: 0.8552
Evaluation Results:
Validation AUC: 0.8552
Validation Evaluation: TPR at 5% FPR = 0.4291, Predictive Equality = 0.9651
Test AUC: 0.8546
Test Evaluation: TPR at 5% FPR = 0.4609, Predictive Equality = 0.9534
Applying RandomOverSampler sampling technique:
Best parameters found: {'n_estimators': 600, 'max_depth': 12, 'min_samples_split': 2, 'max_features': 'log2', 'class_weight': None}
Best validation AUC: 0.8770
Evaluation Results:
Validation AUC: 0.8770
Validation Evaluation: TPR at 5% FPR = 0.4890, Predictive Equality = 0.9879
Test AUC: 0.8685
Test Evaluation: TPR at 5% FPR = 0.4930, Predictive Equality = 0.9890
Applying RandomUnderSampler sampling technique:
Best parameters found: {'n_estimators': 600, 'max_depth': 12, 'min_samples_split': 4, 'max_features': 'log2', 'class_weight': None}
Best validation AUC: 0.8779
Evaluation Results:
Validation AUC: 0.8779
Validation Evaluation: TPR at 5% FPR = 0.5170, Predictive Equality = 1.0000
Test AUC: 0.8834
Test Evaluation: TPR at 5% FPR = 0.5331, Predictive Equality = 1.0000
Applying SVMSMOTE sampling technique:
Best parameters found: {'n_estimators': 600, 'max_depth': 12, 'min_samples_split': 4, 'max_features': 'log2', 'class_weight': None}
Best validation AUC: 0.8576
Evaluation Results:
Validation AUC: 0.8576
Validation Evaluation: TPR at 5% FPR = 0.4311, Predictive Equality = 0.8059
Test AUC: 0.8573
Test Evaluation: TPR at 5% FPR = 0.4729, Predictive Equality = 0.7777
Best parameters found: {'n_estimators': 600, 'max_depth': 10, 'min_samples_split': 2, 'max_features': 'log2', 'class_weight': 'balanced_subsample'}
Best validation AUC: 0.8778
Evaluation Results:
Validation AUC: 0.8778
Validation Evaluation: TPR at 5% FPR = 0.4870, Predictive Equality = 0.9996
Test AUC: 0.8717
Test Evaluation: TPR at 5% FPR = 0.5070, Predictive Equality = 0.9984
Best parameters found: {'n_estimators': 600, 'max_depth': 10, 'min_samples_split': 2, 'max_features': 'log2', 'class_weight': 'balanced'}
Best validation AUC: 0.8773
Evaluation Results:
Validation AUC: 0.8773
Validation Evaluation: TPR at 5% FPR = 0.4970, Predictive Equality = 0.9996
Test AUC: 0.8729
Test Evaluation: TPR at 5% FPR = 0.5050, Predictive Equality = 0.9996
'''
'''
Findings:
RandomUnderSampler extremely good!
'''