-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtrain_final_models_binary.py
216 lines (169 loc) · 11.4 KB
/
train_final_models_binary.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
import numpy as np
import matplotlib.pyplot as plt
from pydantic import Extra
import seaborn as sns
import pandas as pd
from sklearn import svm
from sklearn.preprocessing import StandardScaler, LabelEncoder, OneHotEncoder
from sklearn.model_selection import train_test_split, RepeatedStratifiedKFold, StratifiedKFold
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import f1_score,confusion_matrix, roc_auc_score
from sklearn.metrics import accuracy_score, recall_score
import pickle
import warnings
from sklearn.metrics import classification_report
from sklearn.model_selection import cross_val_predict, cross_val_score
from sklearn.ensemble import ExtraTreesClassifier
import joblib
warnings.filterwarnings('ignore')
### Train Models
def train_model(model, df, df_test, save_model_name, save_scaler_name, age_flag, education_flag, gender_flag, scaling, binary):
### Drop NaN and Inf
df.replace([np.inf, -np.inf], np.nan, inplace=True)
df = df.dropna().reset_index(drop = True)
df_test.replace([np.inf, -np.inf], np.nan, inplace=True)
df_test = df_test.fillna(0).reset_index(drop = True)
dft = df_test
# Drop according to binary
if binary == 'hs':
df['Diagnosis'] = df['Diagnosis'].replace(['E-MCI'],'MCI')
df['Diagnosis'] = df['Diagnosis'].replace(['L-MCI'],'MCI')
df = df[df.Diagnosis != 'MCI']
Y = df.Diagnosis
dft['Diagnosis'] = dft['Diagnosis'].replace(['E-MCI'],'MCI')
dft['Diagnosis'] = dft['Diagnosis'].replace(['L-MCI'],'MCI')
dft = dft[dft.Diagnosis != 'MCI']
Y_test = dft.Diagnosis
elif binary == 'hm':
df['Diagnosis'] = df['Diagnosis'].replace(['E-MCI'],'MCI')
df['Diagnosis'] = df['Diagnosis'].replace(['L-MCI'],'MCI')
df = df[df.Diagnosis != 'SCD']
dft['Diagnosis'] = dft['Diagnosis'].replace(['E-MCI'],'MCI')
dft['Diagnosis'] = dft['Diagnosis'].replace(['L-MCI'],'MCI')
dft = dft[dft.Diagnosis != 'SCD']
Y = df.Diagnosis
Y_test = dft.Diagnosis
elif binary == 'sm':
df['Diagnosis'] = df['Diagnosis'].replace(['E-MCI'],'MCI')
df['Diagnosis'] = df['Diagnosis'].replace(['L-MCI'],'MCI')
df = df[df.Diagnosis != 'Healthy']
dft['Diagnosis'] = dft['Diagnosis'].replace(['E-MCI'],'MCI')
dft['Diagnosis'] = dft['Diagnosis'].replace(['L-MCI'],'MCI')
dft = dft[dft.Diagnosis != 'Healthy']
Y = df.Diagnosis
Y_test = dft.Diagnosis
### Prepare the data-set
print(Y_test.value_counts())
label_1 = LabelEncoder()
if age_flag and education_flag and gender_flag:
X = df[df.columns[~df.columns.isin(['Unnamed: 0','Ratio Q1', 'Name','Diagnosis','Min zcr'])]]
X['Gender']= label_1.fit_transform(X['Gender'])
X['Gender'] = pd.get_dummies(X['Gender'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
X['Education'] = pd.get_dummies(X['Education'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
X_test = dft[dft.columns[~dft.columns.isin(['Unnamed: 0','Ratio Q1', 'Name','Diagnosis','Min zcr'])]]
X_test['Gender']= label_1.transform(X_test['Gender'])
X_test['Gender'] = pd.get_dummies(X_test['Gender'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
X_test['Education'] = pd.get_dummies(X_test['Education'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
s = '_AEG.sav'
elif education_flag and gender_flag:
X = df[df.columns[~df.columns.isin(['Unnamed: 0','Ratio Q1', 'Name','Diagnosis','Min zcr','Age'])]]
X['Gender']= label_1.fit_transform(X['Gender'])
X['Gender'] = pd.get_dummies(X['Gender'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
X['Education'] = pd.get_dummies(X['Education'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
X_test = dft[dft.columns[~dft.columns.isin(['Unnamed: 0','Ratio Q1', 'Name','Diagnosis','Min zcr','Age'])]]
X_test['Gender']= label_1.transform(X_test['Gender'])
X_test['Gender'] = pd.get_dummies(X_test['Gender'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
X_test['Education'] = pd.get_dummies(X_test['Education'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
s = '_EG.sav'
elif gender_flag:
X = df[df.columns[~df.columns.isin(['Unnamed: 0','Ratio Q1', 'Name','Diagnosis','Min zcr','Age','Education'])]]
X['Gender']= label_1.fit_transform(X['Gender'])
X['Gender'] = pd.get_dummies(X['Gender'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
X_test = dft[dft.columns[~dft.columns.isin(['Unnamed: 0','Ratio Q1', 'Name','Diagnosis','Min zcr','Age','Education'])]]
X_test['Gender']= label_1.transform(X_test['Gender'])
X_test['Gender'] = pd.get_dummies(X_test['Gender'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
s = '_G.sav'
else:
X = df[df.columns[~df.columns.isin(['Unnamed: 0','Ratio Q1', 'Name','Diagnosis','Min zcr','Age','Education','Gender'])]]
X_test = dft[dft.columns[~dft.columns.isin(['Unnamed: 0','Ratio Q1', 'Name','Diagnosis','Min zcr','Age','Education','Gender'])]]
s = '.sav'
### Encoding
X['Stress_Depression']= label_1.fit_transform(X['Stress_Depression'])
X['Stress_Depression'] = pd.get_dummies(X['Stress_Depression'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
X_test['Stress_Depression']= label_1.transform(X_test['Stress_Depression'])
X_test['Stress_Depression'] = pd.get_dummies(X_test['Stress_Depression'],prefix_sep='_', dummy_na=False, columns=None,sparse=False, drop_first=False)
X_test_1 = X_test
### Train-test split
x_train, x_test, y_train, y_test = train_test_split(X, Y, test_size=0.2)
### Scaling if need to
if scaling:
scaler = StandardScaler()
x_train = scaler.fit_transform(x_train)
x_test = scaler.transform(x_test)
X = scaler.transform(X)
X_test = scaler.transform(X_test)
joblib.dump(scaler, save_scaler_name)
### Model
clf = model
clf.fit(x_train, y_train)
# feat_importances = pd.Series(clf.feature_importances_, index=X_test_1.columns)
# feat_importances.nlargest(30).plot(kind='barh')
# most_important_feat = feat_importances.nlargest(30).index.tolist()
# plt.show()
### Print model metrics
if binary == 'hs':
print('Classification Report for Train Set: ')
print(classification_report(y_test, clf.predict(x_test), target_names=['Healthy','SCD']))
cv = RepeatedStratifiedKFold(n_splits=10, n_repeats=3)
n_scores = cross_val_score(clf, X, Y, scoring='accuracy', cv=cv, n_jobs=-1, error_score='raise')
print('Cross-Validated Accuracy : %.3f ± (%.3f)' % (np.mean(n_scores), np.std(n_scores)))
print('Classification Report for Test Set: ')
print(Y_test.shape)
print(X_test.shape)
print(classification_report(Y_test, clf.predict(X_test), target_names=['Healthy','SCD']))
elif binary == 'hm':
print('Classification Report for Train Set: ')
print(classification_report(y_test, clf.predict(x_test), target_names=['Healthy','MCI']))
cv = RepeatedStratifiedKFold(n_splits=10, n_repeats=3)
n_scores = cross_val_score(clf, X, Y, scoring='accuracy', cv=cv, n_jobs=-1, error_score='raise')
print('Cross-Validated Accuracy : %.3f ± (%.3f)' % (np.mean(n_scores), np.std(n_scores)))
print('Classification Report for Test Set: ')
print(classification_report(Y_test, clf.predict(X_test), target_names=['Healthy','MCI']))
elif binary == 'sm':
print('Classification Report for Train Set: ')
print(classification_report(y_test, clf.predict(x_test), target_names=['MCI','SCD']))
cv = RepeatedStratifiedKFold(n_splits=10, n_repeats=3)
n_scores = cross_val_score(clf, X, Y, scoring='accuracy', cv=cv, n_jobs=-1, error_score='raise')
print('Cross-Validated Accuracy : %.3f ± (%.3f)' % (np.mean(n_scores), np.std(n_scores)))
print('Classification Report for Test Set: ')
print(classification_report(Y_test, clf.predict(X_test), target_names=['MCI','SCD']))
# Save the model to disk
filename = save_model_name+s
pickle.dump(model, open(filename, 'wb'))
## Load data sets
df1_train = pd.read_csv('df_some_train_1.csv').dropna().reset_index(drop = True)
df2_train = pd.read_csv('df_some_train_2.csv').dropna().reset_index(drop = True)
df3_train = pd.read_csv('df_some_train_3.csv').dropna().reset_index(drop = True)
df4_train = pd.read_csv('df_some_train_4.csv').dropna().reset_index(drop = True)
df5_train = pd.read_csv('df_some_train_5.csv').dropna().reset_index(drop = True)
dfs_train = [df1_train, df2_train, df3_train, df4_train, df5_train]
df1_test = pd.read_csv('df_some_test_1.csv').dropna().reset_index(drop = True)
df2_test = pd.read_csv('df_some_test_2.csv').dropna().reset_index(drop = True)
df3_test = pd.read_csv('df_some_test_3.csv').dropna().reset_index(drop = True)
df4_test = pd.read_csv('df_some_test_4.csv').dropna().reset_index(drop = True)
df5_test = pd.read_csv('df_some_test_5.csv').dropna().reset_index(drop = True)
dfs_test = [df1_test, df2_test, df3_test, df4_test, df5_test]
# models = [svm.SVC(kernel='rbf',C=10,probability=True),svm.SVC(kernel='rbf',C=20,probability=True), svm.SVC(kernel='rbf',C=12,probability=True), svm.SVC(kernel='rbf',C=14,probability=True), svm.SVC(kernel='rbf',C=14,probability=True), svm.SVC(kernel='rbf',C=14,probability=True), svm.SVC(kernel='rbf',C=14,probability=True), svm.SVC(kernel='rbf',C=10,probability=True), svm.SVC(kernel='rbf',C=25,probability=True), svm.SVC(kernel='rbf',C=10,probability=True),]
models = [ExtraTreesClassifier(),ExtraTreesClassifier(),ExtraTreesClassifier(), ExtraTreesClassifier(),ExtraTreesClassifier()]
################ Some people out Testing 5 models:
# model_names = ['SVM_1_hs','SVM_2_hs','SVM_3_hs','SVM_4_hs','SVM_5_hs']
# scaler_names = ['Scaler_1_hs.gz','Scaler_2_hs.gz','Scaler_3_hs.gz','Scaler_4_hs.gz','Scaler_5_hs.gz']
model_names = ['SVM_1_hm','SVM_2_hm','SVM_3_hm','SVM_4_hm','SVM_5_hm']
scaler_names = ['Scaler_1_hm.gz','Scaler_2_hs.gz','Scaler_3_hm.gz','Scaler_4_hm.gz','Scaler_5_hm.gz']
# model_names = ['SVM_1_sm','SVM_2_sm','SVM_3_sm','SVM_4_sm','SVM_5_sm']
# scaler_names = ['Scaler_1_sm.gz','Scaler_2_sm.gz','Scaler_3_sm.gz','Scaler_4_sm.gz','Scaler_5_sm.gz']
for i in range(5):
train_model(models[i], dfs_train[i], dfs_test[i], model_names[i], scaler_names[i], age_flag=False, education_flag=False, gender_flag=False, scaling=True, binary='hm')
# train_model(models[i], dfs_train[i], dfs_test[i], model_names[i], scaler_names[i], age_flag=True, education_flag=True, gender_flag=True, scaling=True)
# train_model(models[i], dfs_train[i], dfs_test[i], model_names[i], scaler_names[i], age_flag=True, education_flag=True, gender_flag=True, scaling=False)
# train_model(models[i], dfs_train[i], dfs_test[i], model_names[i], scaler_names[i], age_flag=False, education_flag=False, gender_flag=False, scaling=False)