-
Notifications
You must be signed in to change notification settings - Fork 0
/
log_reg.py
141 lines (127 loc) · 5.4 KB
/
log_reg.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
#load libraries
import pandas as pd
import numpy as np
import random
from sklearn.metrics import confusion_matrix
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn import preprocessing
from sklearn.decomposition import PCA
import warnings
warnings.filterwarnings("ignore")
def getfscore(cm):
TP=cm[0][0]
TN=cm[1][1]
FN=cm[0][1]
FP=cm[1][0]
precision = TP/(TP+FP)
recall = TP/(TP+FN)
fmesr=(2*precision*recall)/(precision+recall)
return fmesr
#load Dataset
dataset = pd.read_csv("creditcard.csv")
colName = [i for i in dataset.columns]
colName = colName[1:-1]
dataset = dataset.sample(frac = 1) #shuffle
x = dataset.iloc[:,1:-1].values.tolist() #features
y = dataset.iloc[:,-1].values.tolist() #labels
negData = y.count(0)
posData = y.count(1)
negPer = (negData/len(y))*100
posPer = (posData/len(y))*100
print("Fraud data percentage: {0}".format(posPer))
print("Dataset is highly unbalanced. We will have to balance the dataset")
x_neg = dataset.loc[dataset['Class']==0].sample(frac=1)
x_pos = dataset.loc[dataset['Class']==1].sample(frac=1)
#Geneate 5 random datasets
xnegArr=[]
for i in range(5):
k = random.randrange(negData) - posData
xnegArr.append(x_neg[k:k+posData])
new_dsArr=[]
for i in range(5):
new_dsArr.append(pd.concat([x_pos, xnegArr[i]]).sample(frac = 1))
print("Method 2: training using logistic regression....")
#training for one dataset
x_set = new_dsArr[0].iloc[:,1:-1].values.tolist()
x_set_fs= preprocessing.scale(x_set)
y_set = new_dsArr[0].iloc[:,-1].values.tolist()
x_train, x_test, y_train, y_test = train_test_split(x_set, y_set, test_size=0.30, random_state=1)
cls_lr = LogisticRegression(random_state=0,solver='liblinear').fit(x_train, y_train)
y_pred = cls_lr.predict(x_test)
#Confusion matrix
cm = confusion_matrix(y_test, y_pred)
print("For dataset {0}, f-measure score is: {1}".format(1, getfscore(cm)))
#With feature scalling
x_train, x_test, y_train, y_test = train_test_split(x_set_fs, y_set, test_size=0.30, random_state=1)
cls_lr1 = LogisticRegression(random_state=0,solver='liblinear').fit(x_train, y_train)
y_pred = cls_lr1.predict(x_test)
cm = confusion_matrix(y_test, y_pred)
print("With feature scalling, f-measure score is: {0}".format(getfscore(cm)))
print("")
print("using classifier on remaining test dataset with feature scaling")
for i in range(1,5):
x_set = preprocessing.scale(new_dsArr[i].iloc[:,1:-1].values.tolist()) #for logistic Reg
y_set = new_dsArr[i].iloc[:,-1].values.tolist()
y_pred = cls_lr1.predict(x_set)
cm = confusion_matrix(y_set, y_pred)
print("For dataset {0}, f-measure score is: {1}".format(i+1,getfscore(cm)))
print("testing on original dataset")
y_pred = cls_lr1.predict(preprocessing.scale(x))
cm = confusion_matrix(y, y_pred)
print("For original dataset, f-measure score is: {0}".format(getfscore(cm)))
print("")
#PCA for dimentionality reduction
print("Implement PCA for dimentionality reduction...")
pca = PCA(n_components = 15)
pcaArr=[]
for i in range(5):
d1 = pd.DataFrame(pca.fit_transform(preprocessing.scale(new_dsArr[i].iloc[:,1:-1])))
d1.insert(15, "Class", new_dsArr[i].iloc[:,-1].values.tolist(), True)
pcaArr.append(d1)
convArr=np.array(pca.components_)
x_pca = pcaArr[0].iloc[:,0:-1].values.tolist()
y_pca = pcaArr[0].iloc[:,-1].values.tolist()
x_train, x_test, y_train, y_test = train_test_split(x_pca, y_pca, test_size=0.30, random_state=1)
cls_lr_pc = LogisticRegression(random_state=0,solver='liblinear').fit(x_train, y_train)
y_pred = cls_lr_pc.predict(x_test)
#Confusion matrix
cm = confusion_matrix(y_test, y_pred)
print("For dataset {0}, f-measure score is: {1}".format(1, getfscore(cm)))
print("testing on original dataset")
pca_dataframe=[]
x_pre = preprocessing.scale(x)
for i in range(len(x_pre)):
nparr=np.array([x_pre[i]])
nparr=nparr.dot(np.transpose(convArr)).tolist()
pca_dataframe.append([item for sublist in nparr for item in sublist])
pca_dataframe=pd.DataFrame(pca_dataframe)
pca_dataframe.insert(15, "Class", y, True)
X_pca = pca_dataframe.iloc[:,0:-1].values.tolist()
Y_pca = pca_dataframe.iloc[:,-1].values.tolist()
y_pred = cls_lr_pc.predict(X_pca)
cm = confusion_matrix(Y_pca, y_pred)
print("For original dataset, f-measure score is: {0}".format(getfscore(cm)))
print("")
print("weighted Logistic Regression for better output")
ds = pd.concat([x_pos[0:350], x_neg[0:190000]]).sample(frac = 1)
ds_test = pd.concat([x_pos[350:], x_neg[190000:]]).sample(frac = 1)
x_train = preprocessing.scale(ds.iloc[:,1:-1].values.tolist())
y_train = ds.iloc[:,-1].values.tolist()
x_test = preprocessing.scale(ds_test.iloc[:,1:-1].values.tolist())
y_test= ds_test.iloc[:,-1].values.tolist()
#x_train, x_test, y_train, y_test = train_test_split(x_set, y_set, test_size=0.30, random_state=1)
weights = {0:0.1, 1:10} #Initial weight array for weight calculations
cls_lrw = LogisticRegression(solver='saga', class_weight=weights)
cls_lrw.fit(x_train, y_train)
y_pred = cls_lrw.predict(x_train)
cm = confusion_matrix(y_train, y_pred)
print("For train dataset, f-measure score is: {0}".format(getfscore(cm)))
y_pred = cls_lrw.predict(x_test)
cm = confusion_matrix(y_test, y_pred)
print("For test dataset, f-measure score is: {0}".format(getfscore(cm)))
print("")
print("testing on complete dataset")
y_pred = cls_lrw.predict(preprocessing.scale(x))
cm = confusion_matrix(y, y_pred)
print("For original dataset, f-measure score is: {0}".format(getfscore(cm)))