-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathrule_fit.R.R
168 lines (123 loc) · 5.37 KB
/
rule_fit.R.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
library(data.table)
system.time(data<- fread("/home/recsys/zhongziyan/tag_model/tag_49_withskip.txt",sep="\t"))
data <- setNames(data, c("hour","pin","tag","cnt_clk","cnt_ord",
"f1","f2","f3","f4","f5","f6","f7","f8","f9",
"f10","f11","f12","f13","f14","f15","f16","f17","f18","f19",
"f20","f21","f22","f23","f24","f25","f26","f27","f28","f29",
"f30","f31","f32","f33","f34","f35","f36","f37","f38","f39",
"f40","f41","f42","f43","f44","f45","f46","f47","f48","skip_real"))
data$prc_grade_diff_bin=0
data$prc_grade_diff_bin[data$prc_grade_diff!=0]=1
random <- runif(nrow(data), 0, 1)
data1 <- data[order( random), ]
trainset <- data1[1:ceiling(nrow(data1)/2), ]
testset<- data1[(ceiling(nrow(data1)/2) + 1) : nrow(data1), ]
library(gbm)
model <- gbm(cvr~., data=trainset,shrinkage=0.05,
cv.folds=4,n.trees=300,verbose=F)
best.iter <- gbm.perf(model,method='cv')
summary(model,best.iter)
xx_train <- trainset[, c(1,3,6:53),with=F]
yy <- trainset[,skip_real]
xx_test <- testset[, c(1,3,6:53),with=F]
yytest <- testset[,skip_real]
f <- glm(yy ~ ., data = xx_train, family = 'binomial')
summary(f)
glm_pre=predict(f, xx_test)
glm_pre=as.numeric(glm_pre)
aa=cbind(glm_pre,yytest)
aa1=aa[yytest>0,]
seaa=mean(abs(aa1[,1]-aa1[,2]))
se=mean(abs(glm_pre-yytest))
library(glmnet)
glm.net.sol2 <- cv.glmnet(xx_train, yy, alpha = 0.5)
coef2<- coef(glm.net.sol2, s=glm.net.sol2$lambda.min)
system.time(glm.net.sol3<- cv.glmnet(as.matrix(xx_train), yy, alpha = 0.1,intercept=FALSE,type.measure='auc',family="binomial"))
coef3<- coef(glm.net.sol3, s=glm.net.sol3$lambda.min)
glm.net.sol4<- cv.glmnet(xx_train, yy, alpha = 0.01,intercept=FALSE,lower.limits=0)
coef4<- coef(glm.net.sol4, s=glm.net.sol4$lambda.min)
system.time(glm.net.sol5<- cv.glmnet(xx_train[,c(-19,-20)], yy, alpha = 0.01,intercept=FALSE))
coef5<- coef(glm.net.sol5, s=glm.net.sol5$lambda.min)
print(glm.net.sol3)
glm.net.sol1<- cv.glmnet(xx_train, yy, alpha = 0.5,lower.limits=0)
plot(glm.net.sol2 ,xvar="lambda",label=TRUE)
pred <- predict(glm.net.sol3, newx=xx_test, s=glm.net.sol3$lambda.min)
pred=as.numeric(pred)
aa=cbind(pred,yytest)
aa1=aa[yytest>0,]
seaa=mean(abs(aa1[,1]-aa1[,2]))
se=mean(abs(pred-yytest))
pred1 <- predict(glm.net.sol2, newx=xx_test, s=glm.net.sol2$lambda.min)
pred1=as.numeric(pred1)
aa=cbind(pred1,yytest)
aa1=aa[yytest>0,]
seaa1=mean(abs(aa1[,1]-aa1[,2]))
se1=mean(abs(pred1-yytest))
pred2 <- predict(glm.net.sol4, newx=xx_test, s=glm.net.sol4$lambda.min)
pred2=as.numeric(pred2)
aa=cbind(pred2,yytest)
aa1=aa[yytest>0,]
seaa2=mean(abs(aa1[,1]-aa1[,2]))
se2=mean(abs(pred2-yytest))
pred3 <- predict(glm.net.sol5, newx=xx_test[,c(-26,-27,-35)], s=glm.net.sol5$lambda.min)
pred3=as.numeric(pred3)
aa=cbind(pred3,yytest)
aa1=aa[yytest>0,]
seaa3=mean(abs(aa1[,1]-aa1[,2]))
se3=mean(abs(pred3-yytest))
trainset <- data1[1:ceiling(nrow(data1)/8), ]
testset<- data1[(ceiling(nrow(data1)/8) + 1) : nrow(data1), ]
Coefficients: (2 not defined because of singularities)
Estimate Std. Error z value Pr(>|z|)
(Intercept) -1.715e+02 7.535e+00 -22.766 < 2e-16 ***
hour 8.645e-03 4.024e-04 21.480 < 2e-16 ***
f2 -1.008e-01 1.220e-02 -8.265 < 2e-16 ***
f6 -3.675e-01 6.530e-03 -56.285 < 2e-16 ***
f9 2.485e-01 2.323e-03 106.992 < 2e-16 ***
f13 5.986e+01 2.574e+00 23.257 < 2e-16 ***
f21 -5.099e+02 2.195e+01 -23.228 < 2e-16 ***
f23 -2.855e+02 4.108e+01 -6.951 3.64e-12 ***
f47 NA NA NA NA
f48 NA NA NA NA
platform = "linux"
rfhome ="/home/recsys/xiongxi/rulefit"
source("/home/recsys/xiongxi/rulefit/rulefit.r")
install.packages("akima", lib=rfhome)
library(akima, lib.loc=rfhome)
yy[yy<1]=-1
rfmod = rulefit (xx_train, yy, cat.vars=c(1,2), not.used=c(49),rfmode="class", sparse=0.5, test.reps=0, test.fract=0.2, mod.sel=1, model.type="both", tree.size=4, max.rules=200 , max.trms=50 , costs=c(2,1), quiet=F)
runstats(rfmod)
vi = varimp (impord =TRUE,x=xx_train)
rules( beg=1, end=10, x=xx_train )
D:/RecModels/rec130000/globle mercury/rulefit
rules=function(beg=1,end=beg+9,x=NULL,wt=rep(1,n)) {
beg <- parchk("beg",beg,1,1000000,1)
end <- parchk("end",end,beg,1000000,beg+9)
zz=file(paste(rfhome,'/intparms',sep=''),'rb')
ip=readBin(zz,integer(),size=4,n=4); close(zz); ni=ip[4]
if(!missing(x)) { mode=1
if(is.data.frame(x)) { p=length(x); n=length(x[[1]])
xx=matrix(nrow=n,ncol=p); for (j in 1:p) { xx[,j]=x[[j]]}
}
else if(is.matrix(x)) { n <- nrow(x); p <- ncol(x); xx=x;}
else if(is.vector(x)) { p=length(x); n=1; xx=x}
else { stop(' x must be a data frame, matrix, or vecing message:
running command 'cmd /c rf_go.exe' had status 41 tor.')}
if(p!=ni)
stop(paste(' number of variables =',as.character(p),
'is different from training data =',as.character(ni)))
if(any(is.na(xx))) {
zz=file(paste(rfhome,'/realparms',sep=''),'rb')
xmiss=readBin(zz,numeric(),size=8,n=1); close(zz)
xx[is.na(xx)] <- xmiss;
}
zz=file(paste(rfhome,'/varimp.x',sep=''),'wb')
writeBin(as.numeric(c(n,as.vector(xx))),zz,size=8); close(zz)
putw(wt,xx,n,p,'varimp.w')
}
else { mode=0}
zz=file(paste(rfhome,'/intrules',sep=''),'wb')
writeBin(as.integer(c(mode,beg,end)),zz,size=4); close(zz)
wd=getwd(); setwd(rfhome)
status=rfexe('rules')
}