-
Notifications
You must be signed in to change notification settings - Fork 0
/
ML.py
143 lines (118 loc) · 5.47 KB
/
ML.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
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import pickle
from sklearn.metrics import roc_auc_score
from sklearn.model_selection import GridSearchCV
from sklearn.model_selection import train_test_split
from sklearn.model_selection import cross_val_score
from sklearn.preprocessing import StandardScaler , MinMaxScaler,MaxAbsScaler
from sklearn.preprocessing import Binarizer
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score
from sklearn.metrics import confusion_matrix
from sklearn.model_selection import StratifiedKFold
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier
from sklearn.preprocessing import PowerTransformer
from imblearn.over_sampling import SMOTE
from collinearity import SelectNonCollinear
from sklearn.feature_selection import VarianceThreshold
from lightgbm import LGBMClassifier
from sklearn.impute import SimpleImputer
#Seperate Feature
data = pd.read_csv('wafer_data.csv')
y = data["Class"]
x = data.drop(['Class'], axis = 1)
#StandardScaler
SS = StandardScaler()
SS.fit(x)
X_SS = SS.transform(x)
X_SS_PD = pd.DataFrame(X_SS)
#MaxAbsScaler
MAS = MaxAbsScaler()
MAS.fit(x)
X_MAS = MAS.transform(x)
#yeo-johnson Transformation
PTF=PowerTransformer()
PTF.fit(X_MAS)
X_PTF = PTF.transform(X_MAS)
#remove_multicollinearity
selector = SelectNonCollinear(0.90)
selector.fit(X_PTF,y)
X_sel=selector.transform(X_PTF)
#feature_selection
VT = VarianceThreshold(threshold=(.3 * (1 - .3)))
XXXX=VT.fit_transform(X_sel)
#unknown_categorical
Xi = SimpleImputer(strategy='mean')
XX=Xi.fit_transform(XXXX)
#most_frequent
X_train1, X_test1, Y_train1, Y_test1 = train_test_split(XX, y, test_size=0.2, random_state=10)
X_train2, X_test2, Y_train2, Y_test2 = train_test_split(XX, y, test_size=0.2, random_state=20)
X_train3, X_test3, Y_train3, Y_test3 = train_test_split(XX, y, test_size=0.2, random_state=30)
X_train4, X_test4, Y_train4, Y_test4 = train_test_split(XX, y, test_size=0.2, random_state=40)
X_train5, X_test5, Y_train5, Y_test5 = train_test_split(XX, y, test_size=0.2, random_state=50)
X_train6, X_test6, Y_train6, Y_test6 = train_test_split(XX, y, test_size=0.2, random_state=60)
X_train7, X_test7, Y_train7, Y_test7 = train_test_split(XX, y, test_size=0.2, random_state=70)
X_train8, X_test8, Y_train8, Y_test8 = train_test_split(XX, y, test_size=0.2, random_state=80)
X_train9, X_test9, Y_train9, Y_test9 = train_test_split(XX, y, test_size=0.2, random_state=90)
X_train10, X_test10, Y_train10, Y_test10 = train_test_split(XX, y, test_size=0.2, random_state=100)
#fix_imbalance
smote = SMOTE(random_state=10)
X_train_over1,y_train_over1 = smote.fit_resample(X_train1,Y_train1)
X_train_over2,y_train_over2 = smote.fit_resample(X_train2,Y_train2)
X_train_over3,y_train_over3 = smote.fit_resample(X_train3,Y_train3)
X_train_over4,y_train_over4 = smote.fit_resample(X_train4,Y_train4)
X_train_over5,y_train_over5 = smote.fit_resample(X_train5,Y_train5)
X_train_over6,y_train_over6 = smote.fit_resample(X_train6,Y_train6)
X_train_over7,y_train_over7 = smote.fit_resample(X_train7,Y_train7)
X_train_over8,y_train_over8 = smote.fit_resample(X_train8,Y_train8)
X_train_over9,y_train_over9 = smote.fit_resample(X_train9,Y_train9)
X_train_over10,y_train_over10 = smote.fit_resample(X_train10,Y_train10)
#rf = LGBMClassifier(n_estimators=100,min_split_gain=0.9,min_child_samples=21,learning_rate=0.01,num_leaves=40,reg_alpha=0.005,reg_lambda=0.0005,subsample_for_bin=200000,feature_fraction=0.6,bagging_freq=6,bagging_fraction=0.6)
#min_split_gain=0.9,min_child_samples=21,learning_rate=0.01,num_leaves=40,reg_alpha=0.005,reg_lambda=0.0005,subsample_for_bin=200000,feature_fraction=0.6,bagging_freq=6,bagging_fraction=0.6
rf= RandomForestClassifier(bootstrap=False, n_estimators=120,max_features='sqrt',min_impurity_decrease =0.0005)
#criterion='entropy'
#max_depth=10
#max_features='sqrt',min_samples_leaf=4,min_samples_split=2,,min_impurity_decrease =0.0005
rf.fit(X_train_over1,y_train_over1)
rf.fit(X_train_over2,y_train_over2)
rf.fit(X_train_over3,y_train_over3)
rf.fit(X_train_over4,y_train_over4)
rf.fit(X_train_over5,y_train_over5)
rf.fit(X_train_over6,y_train_over6)
rf.fit(X_train_over7,y_train_over7)
rf.fit(X_train_over8,y_train_over8)
rf.fit(X_train_over9,y_train_over9)
rf.fit(X_train_over10,y_train_over10)
# Accuracy Check!!
pred1 = rf.predict(X_test1)
pred2 = rf.predict(X_test2)
pred3 = rf.predict(X_test3)
pred4 = rf.predict(X_test4)
pred5 = rf.predict(X_test5)
pred6 = rf.predict(X_test6)
pred7 = rf.predict(X_test7)
pred8 = rf.predict(X_test8)
pred9 = rf.predict(X_test9)
pred10 = rf.predict(X_test10)
# pred_t = rf.predict(X_train_over)
# print(accuracy_score(Y_test, pred))
# print(accuracy_score(y_train_over,pred_t))
# print(confusion_matrix(y_train_over, pred_t))
AUC_AVG=roc_auc_score(Y_test1, pred1)+roc_auc_score(Y_test2, pred2)+roc_auc_score(Y_test3, pred3)+roc_auc_score(Y_test4, pred4)+roc_auc_score(Y_test5, pred5)+roc_auc_score(Y_test6, pred6)+roc_auc_score(Y_test7, pred7)+roc_auc_score(Y_test8, pred8)+roc_auc_score(Y_test9, pred9)+roc_auc_score(Y_test10, pred10)
print(AUC_AVG/10)
#sns.boxplot(data=X_SS)
#plt.show()
'''
#Use VAL_Score
log_reg_kf = LogisticRegression(random_state=13, solver = 'liblinear')
#log_reg_kf = RandomForestClassifier(n_estimators=2000)
skfold = StratifiedKFold(n_splits = 5)
score = cross_val_score(log_reg_kf, X_train, Y_train,scoring='accuracy', cv = 5)
print(np.mean(score))
with open('model_face.pkl','wb') as f:
pickle.dump(rf,f,protocol=2)
'''