Machine-Learning/Checkpoint#3.R at master · rahulsakore7/Machine-Learning · GitHub

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
#setwd('C:/Users/Student/workspace')

rm(list=ls())
dat <- read.csv("/Users/Dhan/Downloads/Consumer_Complaints.csv", header = TRUE)
headers=names(dat)
headers

# Company response to consumer analysis

in_progress <- which(sapply(dat[15], function(x) (x=="In progress")))
untimely <- which(sapply(dat[15], function(x) (x=="Untimely response")))

head(dat[17])
dat <- dat[-c(in_progress,untimely),]
##650807


summary(dat)

#Company.response.to.consumer
#Closed with explanation        :484563
#Closed with non-monetary relief: 83332
#Closed with monetary relief    : 44460
#Closed without relief          : 17863
#Closed                         : 15291
#Closed with relief             :  5298
#(Other)                        :     0


levels(dat$Company.response.to.consumer)
summary(dat$Company.response.to.consumer)

levels(dat$Company.response.to.consumer)[levels(dat$Company.response.to.consumer) == "Closed with explanation"] <- "No"
levels(dat$Company.response.to.consumer)[levels(dat$Company.response.to.consumer) == "Closed without relief"] <- "No"
levels(dat$Company.response.to.consumer)[levels(dat$Company.response.to.consumer) == "Closed"] <- "No"


levels(dat$Company.response.to.consumer)[levels(dat$Company.response.to.consumer) == "Closed with monetary relief"] <- "Yes"
levels(dat$Company.response.to.consumer)[levels(dat$Company.response.to.consumer) == "Closed with non-monetary relief"] <- "Yes"
levels(dat$Company.response.to.consumer)[levels(dat$Company.response.to.consumer) == "Closed with relief"] <- "Yes"


levels(dat$Company.response.to.consumer)
summary(dat$Company.response.to.consumer)

#### Removed other levels, now only NO and YES are left
dat$Company.response.to.consumer <- factor(dat$Company.response.to.consumer)

table(dat$Company.response.to.consumer)
# Comparing likelihood of response"Yes" from the company
Uniform_prior <- table(dat[15])[2]/(table(dat[15])[1] + table(dat[15])[2])
Uniform_prior
summary(dat[15])


#table(dat[15])[1]+ table(dat[15])[2]
# Creating a table with company and response for dat
j<-with(dat,table(Company.response.to.consumer,Company))
#print(j)
dim(j)
# 2 3890
j1<-j[1,]+j[2,]
print(j1)
j2<-j1>25
# Gives a boolean value for the given expression(Ans:[1]   2 908)
print(j2)
j3<-j[,j2]
print(j3[2,2])
#dim(j3)
chisqv<-array(0,dim =c(dim(j3)[2],2))
for (i in 1:dim(j3)[2]) {
  T=j3[1,i]+j3[2,i]
  Exp_val=T*Uniform_prior
  Yes=j3[2,i]
  chisqvalue=(Yes-Exp_val)^2/Exp_val
  chisqv[i,1]=1- pchisq(chisqvalue,1)
  chisqv[i,2]=(Yes/T)/Uniform_prior

}
rownames(chisqv)<-names(j3[1,])

lift_sort<-sort(chisqv[,2],decreasing = TRUE,index.return=TRUE)$ix
pval_sort<-sort(chisqv[,1],decreasing = TRUE,index.return=TRUE)$ix
lift_value<-chisqv[lift_sort,]
p_value<-chisqv[pval_sort,]
prob_fraud<-names(lift_value[,1])
j[,prob_fraud]
lift_list<-cbind(lift_value,array(0,dim = c(908,2)))
lift_list[,3:4]<-t(j[,prob_fraud])
print(lift_list)

# User experiences

Uniform_prior1 <- table(dat[17])[2]/(table(dat[17])[1] + table(dat[17])[2])
Uniform_prior1
summary(dat[15])


#table(dat[15])[1]+ table(dat[15])[2]
# Creating a table with company and response for dat
jj<-with(dat,table(Product,Consumer.disputed.))
h<-chisq.test(jj)
h
print(jj)
dim(jj)

chisqv1<-array(0,dim =c(dim(jj)[1],2))
for (i in 1:dim(jj)[1]) {
  T1=jj[i,1]+jj[i,2]
  Exp_val1=T1*Uniform_prior1
  Yes=jj[i,2]
  chisqvalue1=(Yes-Exp_val1)^2/Exp_val1
  chisqv1[i,1]=1- pchisq(chisqvalue1,1)
  chisqv1[i,2]=(Yes/T1)/Uniform_prior1

}
rownames(chisqv1)<-names(jj[,1])


#pval_sort<-sort(chisqv[,1],decreasing = TRUE,index.return=TRUE)$ix
lift_value<-chisqv1
lift_value
lift_list1<-cbind(lift_value,array(0,dim=dim(lift_value)))
dim(lift_list1)

lift_list1[,3:4]<- jj[,2:3]
print(lift_list1)
lift_sort<-sort(chisqv1[,2],decreasing = TRUE,index.return=TRUE)$ix
lift_list1<-lift_list1[lift_sort,]
#Predictor builder
yesrows=which(dat[15]=="Yes")
norows=which(dat[15]=="No")
rows=c(yesrows[1:5000],norows[1:5000])
rows1
dim(rows1)
rm(rows1)
outcome_ind=15
toInclude=c(2,3,4,5,9,13)
testdat<-dat[rows,toInclude]
dim(testdat)
#head(testdat[6])
#head(dat[17])
#names(dat)
outcome<-dat[rows,outcome_ind]
#rm(array)
array1<-model.matrix(~.,data = testdat[1])
for (j in 2:length(toInclude)) {
  model<-model.matrix(~.,data = testdat[j])
  array2<-cbind(array1,model)

}

library(randomForest)
forest<-randomForest(x=array2,y=outcome)

forest$confusion
forest$proximity
forest$err.rate
forest$oob.times
forest$ntree
forest$mtry