-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathR-gbmtest-3.r
142 lines (113 loc) · 3.6 KB
/
R-gbmtest-3.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
rm(list=ls())
gc()
sprune<-function(fvar,levs=NA,supp=5000) {
if (length(levs)<=1) {
xx<-table(fvar)
levs<-row.names(xx)[which(xx>supp)]
}
fvarn<-rep("Others",length(fvar))
for (ll in levs) {
fvarn[which(fvar==ll)]<-ll
}
fvarn<-as.factor(as.character(fvarn))
fvarn
}
#library(randomForest)
#This takes a while
train_dat<-read.csv("train.csv")
test_dat<-read.csv("test.csv")
trainlabels_dat<-read.csv("trainLabels.csv")
train_dat[is.na(train_dat)]<-"None"
test_dat[is.na(test_dat)]<-"None"
ldvars<-c("x3","x4","x34","x35","x61","x64","x65","x91","x94","x95")
# for (ldvar in ldvars) {
# train_dat[,ldvar]<-sprune(train_dat[,ldvar],supp=5000)
# test_dat[,ldvar]<-sprune(test_dat[,ldvar],levs=levels(train_dat[,ldvar]))
# }
for (ldvar in ldvars) {
train_dat[,ldvar]<-sprune(train_dat[,ldvar],supp=7500)
test_dat[,ldvar]<-sprune(test_dat[,ldvar],levs=levels(train_dat[,ldvar]))
}
for (ldvar in ldvars) { print(c(ldvar,length(levels(train_dat[,ldvar])),length(levels(test_dat[,ldvar]))))}
trainlabels_dat<-as.data.frame(apply(trainlabels_dat,2,as.factor))
train_index<-sample(1:nrow(train_dat),0.8*nrow(train_dat))
test_index<-setdiff(1:nrow(train_dat),train_index)
rf<-list()
set.seed(123)
x=train_dat[train_index,2:ncol(train_dat)]
xtest=train_dat[test_index,2:ncol(train_dat)]
stestx<-test_dat[,2:ncol(test_dat)]
# for(i in 1:1){
# y=trainlabels_dat[train_index,i+1]
# ytest=trainlabels_dat[test_index,i+1]
# nmin=sum(trainlabels_dat[train_index,i+1]==1)
# rf[[i]]<-randomForest(x=x,y=y,xtest=xtest,ytest=ytest,ntree=20,sampsize=c(nmin,nmin),strata=y,do.trace=TRUE,nodesize=10,keep.forest=TRUE)
# }
rm(list=c("train_dat"))
gc()
library(caret)
# fitControl <- trainControl(## 5-fold CV
# method = "repeatedcv",
# number = 1,
# repeated 3 times
# repeats = 1)
fitControl <- trainControl(## OOB
method = "cv",p=0.9)
gbmGrid <- expand.grid(interaction.depth = c(5,10),
n.trees = c(5,10,15)*10,
shrinkage = 0.1)
nrow(gbmGrid)
#gbmFit2<-list()
trainpreds<-list()
testpreds<-list()
stestprobs<-list()
cmtr<-list()
cmte<-list()
for(i in 1:33) {
print(paste(c("=================",i,"=====================")));
y=trainlabels_dat[train_index,i+1]
ytest=trainlabels_dat[test_index,i+1]
#Undersampling of majority class except y33
set.seed(123)
if (i!=14) {
y1samp<-which(y=='1')
if (length(y1samp)>50000) {
y1samp<-sample(y1samp,50000,replace=T)
}
y0samp<-sample(which(y=='0'),length(y1samp),replace=T)
xn<-x[c(y0samp,y1samp),]
yn<-y[c(y0samp,y1samp)]
set.seed(825)
gbmFit2 <- train(x = xn, y=yn,
method = "gbm",
trControl = fitControl,
verbose = TRUE,
## Now specify the exact models
## to evaludate:
tuneGrid = gbmGrid)
trainpreds[[i]]<-predict(gbmFit2,newdata=xn)
testpreds[[i]]<-predict(gbmFit2,newdata=xtest)
cmtr[[i]]<-confusionMatrix(trainpreds[[i]],yn)
cmte[[i]]<-confusionMatrix(testpreds[[i]],ytest)
stestprobs[[i]]<-extractProb(list(gbm=gbmFit2),unkX=stestx)$X1
} else {
stestprobs[[i]]<-rep(0.0,nrow(stestx))
}
}
for (i in 1:33) {print(cmtr[[i]]$byClass)}
for (i in 1:33) {print(cmte[[i]]$byClass)}
for (i in 1:33) {
print (i)
print (summary(stestprobs[[i]]))
}
save(stestprobs,file="R-gbm-samp-try3.Rdata")
##Writing output
res<-read.csv("sampleSubmission.csv")
nrespt<-nrow(res)/33
#for (j in 1:nrespt) {
for (i in 1:33) {
res$pred[i+33*(0:(nrespt-1))]<-stestprobs[[i]][1:nrespt]
}
#}
summary(res)
write.csv(res,"R-gbm-samp-try3.csv",row.names=FALSE)