1
1

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?

More than 5 years have passed since last update.

bootstrap 事始め 

Last updated at Posted at 2014-10-20

この本の第一章例3をRで。

データの定義

x<- c(5,10,15,20,25,30,35,40,45,50)
y<- c(8.6,10.4,21.6,22.8,19.5,30.0,29.5,34.3,38.7,47.4)

線形回帰

re<-lm(y~x)

その残差

resi<-summary(re)$residual

箱を作って

ncount<-100000
Res<-matrix(0,ncount,length(x))
res.boot<-matrix(0,ncount,length(x))
result<-matrix(0,ncount,2)

一応時間なんかも測ったりして

time<-proc.time()

R初心者です(ループ)

for(i in 1:ncount){
Res[i,]<-sample(resi,replace=T)
res.boot[i,]<-re$coefficients[1]*array(1,length(x))+re$coefficients[2]*x+Res[i,]
temp<-lm(res.boot[i,]~x)
result[i,1]<-temp$coefficients[1]
result[i,2]<-temp$coefficients[2]
}
時間見て
proc.time()-time

ユーザ システム 経過(秒)
1.16 0.00 1.16

1000回で1.5秒か・・・。

summary(result[,1])

Min. 1st Qu. Median Mean 3rd Qu. Max.
-0.8874 3.6070 4.9510 4.9550 6.3630 11.0900

sd(result[,1])
[1] 2.070965

ちなみにただの線形回帰なら

re$coefficients[1]
(Intercept)
4.9

点推定値はいいけど結構散らばっている。まあサンプル10個だからな。

summary(result[,2])

Min. 1st Qu. Median Mean 3rd Qu. Max.
0.6035 0.7282 0.7741 0.7757 0.8175 0.9668

sd(result[,2])
[1] 0.06594192

回帰係数は

re$coefficients[2]
x
0.7774545

こっちはほとんど動かない。
パッケージがあるかとかは調べてない。

一応これでブートストラップ。
事始めでもないか。一応回帰だし。

1
1
0

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
1
1

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?