-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathcredit_card_default.R
187 lines (140 loc) · 5.19 KB
/
credit_card_default.R
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
install.packages('caTools')
install.packages('data.table')
install.packages('mlbench')
install.packages('randomForest')
install.packages('caret')
install.packages('class')
install.packages('tree')
install.packages('ggplot2')
install.packages('ipred')
library(ipred)
library(class)
library(randomForest)
library(tree)
library(Amelia)
library(caTools)
library(readxl)
library(caret)
#import dataset
default_of_credit_card_clients <- read_excel("default of credit card clients.xls")
credit_default <- default_of_credit_card_clients
View(credit_default)
#rename for better visualizations
credit_default$AGE<-cut(credit_default$AGE, breaks = c( 10,30,50,100), labels = c("youth", "middle-aged","elderly"))
credit_default$SEX<-cut(credit_default$SEX, 2,labels = c("F","M"))
credit_default$MARRIAGE<-cut(credit_default$MARRIAGE, 4,labels = c("Married","Single","Divorce","Other"))
#convert integers to factors
credit_default$SEX <- as.factor(credit_default$SEX)
credit_default$EDUCATION <- as.factor(credit_default$EDUCATION)
credit_default$MARRIAGE <- as.factor(credit_default$MARRIAGE)
credit_default$PAY_0 <- as.factor(credit_default$PAY_0)
credit_default$PAY_2 <- as.factor(credit_default$PAY_2)
credit_default$PAY_3 <- as.factor(credit_default$PAY_3)
credit_default$PAY_4 <- as.factor(credit_default$PAY_4)
credit_default$PAY_5 <- as.factor(credit_default$PAY_5)
credit_default$PAY_6 <- as.factor(credit_default$PAY_6)
credit_default$default <- as.factor(credit_default$`default payment next month`)
credit_default <- credit_default[,-c(1,25)]
# ggplot boxplot for age ,limit bal and default
ggplot(data=credit_default,mapping = aes(x=AGE,y=credit_default$LIMIT_BAL,fill=default)) + geom_boxplot()
# bar plot
ggplot(credit_default, mapping = aes(x=MARRIAGE, fill=default)) + geom_bar()
#check for missing values
sapply(credit_default,function(x) sum(is.na(x)))
#Visual representation of missing data
missmap(credit_default, main = "Missing values vs observed")
#plot data imbalance
ggplot(data = credit_default) +
geom_bar(mapping = aes(x = default))
View(credit_default)
#feature selection
#variable importance
fit = randomForest(default~., data=credit_default)
varImpPlot(fit,type=2)
knn_data <- credit_default[,c(6,12,13,14,15,17,24)]
knn_data$PAY_0 <- as.numeric(knn_data$PAY_0)
str(knn_data)
#function to normalize data
normalize <- function(x){
return ((x-min(x)) / (max(x) - min(x)))
}
#split data
set.seed(111)
split = sample.split(knn_data, SplitRatio = 0.75)
#train data
knn_train = subset(knn_data, split == TRUE)
#upsample
set.seed(100)
'%ni%' <- Negate('%in%')
up_train <- upSample(x = knn_train[, colnames(knn_train) %ni% "default"],
y = knn_train$default)
#down sample
set.seed(100)
'%ni%' <- Negate('%in%')
down_train <- downSample(x = knn_train[, colnames(knn_train) %ni% "default"],
y = knn_train$default)
#test data
knn_test = subset(knn_data, split == FALSE)
#use KNN to run knn for up sampled data
train_normalized <- as.data.frame(lapply(up_train[,-7], normalize))
test_normalized <- as.data.frame(lapply(knn_test[,-7], normalize))
k_mod <- knn(train = train_normalized, test = test_normalized, cl = up_train$Class , k=159, prob=TRUE)
#check confusion matrix
k_conf <- confusionMatrix(k_mod, knn_test$default, positive = '1')
#use KNN to run knn for down sampled data using caret and CV
set.seed(123)
down_model <- train(
Class ~., data = down_train, method = "knn",
trControl = trainControl("cv", number = 10),
preProcess = c("center","scale"),
tuneLength = 200
)
#use model to predict
knn_predictions <- predict(down_model, knn_test[,-7])
#check confusion matrix
down_conf <- confusionMatrix(knn_test$default, knn_predictions, positive = '1')
#decision tree with all the variables
#split data for decision tree
set.seed(123)
split = sample.split(credit_default, SplitRatio = 0.75)
#train data
credit_train = subset(credit_default, split == TRUE)
#upsample
set.seed(100)
'%ni%' <- Negate('%in%')
upsample_train <- upSample(x = credit_train[, colnames(credit_train) %ni% "default"],
y = credit_train$default)
#test data
credit_test = subset(credit_default, split == FALSE)
View(credit_test)
#build decision tree
tree.default =tree(Class~., upsample_train)
summary(tree.default)
plot(tree.default)
text(tree.default, pretty = 0)
tree.default
tree.pred =predict(tree.default,credit_test[,-24],type ="class")
#check confusion matrix
confusionMatrix(tree.pred, credit_test$default, positive = '1')
#prune tree
cv.default =cv.tree(tree.default ,FUN=prune.misclass)
names(cv.default )
cv.default
par(mfrow =c(1,2))
plot(cv.default$size ,cv.default$dev ,type="b")
plot(cv.default$k ,cv.default$dev ,type="b")
#decision tree with only the variables used for KNN
tree.data =tree(Class~., up_train)
summary(tree.data)
plot(tree.data)
text(tree.data, pretty = 0)
tree.data
tree.data_pred =predict(tree.data,knn_test[,-7],type ="class")
#confusion matrix
confusionMatrix(tree.data_pred, knn_test$default, positive = '1')
#bagged decision tree
dec_bag <- bagging(Class~., data= up_train, nbagg=25)
#predict with bagged decision tree
dec_bag_pred <- predict(dec_bag, knn_test[,-7], type="class")
#confusion matrix
confusionMatrix(dec_bag_pred, knn_test$default, positive = "1")