plsRglm (version 0.3.3)

bootpls: Non-parametric Bootstrap for PLS models

Description

Provides a wrapper for the bootstrap function boot from the boot R package. Implements non-parametric bootstrap for PLS generalized linear models by case resampling.

Usage

bootpls(object, typeboot="plsmodel", R=250, statistic=coefs.plsR, sim="ordinary", stype="i", ...)

Arguments

object

~~Explain object here~~

typeboot

~~Explain typeboot here~~

R

~~Explain R here~~

statistic

~~Explain statistic here~~

sim

~~Explain sim here~~

stype

~~Explain stype here~~

...

~~Explain ... here~~

Value

• ~Describe the value returned If it is a LIST, use
• comp1
  Description of 'comp1'
• comp2
  Description of 'comp2'
• ...

Details

~~ More details than the description above ~~

References

Lazraq et al. Bastien et al. Davison et al.

See Also

boot

Examples

data(Cornell)
XCornell<-Cornell[,1:7]
yCornell<-Cornell[,8]

# Lazraq-Cl�roux PLS ordinary bootstrap

set.seed(250)
Cornell.boot <- bootpls(plsR(yCornell,XCornell,3), sim="ordinary", stype="i", R=250)
boot::boot.array(Cornell.boot, indices=TRUE)

# Graph similar to the one of Bastien et al. in CSDA 2005
boxplot(as.vector(Cornell.boot$t[,-1])~factor(rep(1:7,rep(250,7))), main="Bootstrap distributions of standardised bj (j = 1, ..., 7).")
points(c(1:7),Cornell.boot$t0[-1],col="red",pch=19)
# Using the boxplots.bootpls function
boxplots.bootpls(Cornell.boot,indices=2:8)
# Confidence intervals plotting
confints.bootpls(Cornell.boot,indices=2:8)
plots.confints.bootpls(confints.bootpls(Cornell.boot,indices=2:8))


library(boot)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=2)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=3)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=4)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=5)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=6)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=7)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=8)

plot(Cornell.boot,index=2)
boot::jack.after.boot(Cornell.boot, index=2, useJ=TRUE, nt=3)
plot(Cornell.boot,index=2,jack=TRUE)

car::data.ellipse(Cornell.boot$t[,2], Cornell.boot$t[,3], cex=.3, levels=c(.5, .95, .99), robust=T)

rm(Cornell.boot)


# PLS balanced bootstrap

set.seed(225)
Cornell.boot <- bootpls(plsR(yCornell,XCornell,3), sim="balanced", stype="i", R=250)
boot::boot.array(Cornell.boot, indices=TRUE)


# Graph similar to the one of Bastien et al. in CSDA 2005
boxplot(as.vector(Cornell.boot$t[,-1])~factor(rep(1:7,rep(250,7))), main="Bootstrap distributions of standardised bj (j = 1, ..., 7).")
points(c(1:7),Cornell.boot$t0[-1],col="red",pch=19)
# Using the boxplots.bootpls function
boxplots.bootpls(Cornell.boot,indices=2:8)
# Confidence intervals plotting
confints.bootpls(Cornell.boot,indices=2:8)
plots.confints.bootpls(confints.bootpls(Cornell.boot,indices=2:8))


library(boot)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=2)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=3)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=4)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=5)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=6)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=7)
boot::boot.ci(Cornell.boot, conf = c(0.90,0.95), type = c("norm","basic","perc","bca"), index=8)

plot(Cornell.boot,index=2)
boot::jack.after.boot(Cornell.boot, index=2, useJ=TRUE, nt=3)
plot(Cornell.boot,index=2,jack=TRUE)

rm(Cornell.boot)

# PLS permutation bootstrap

set.seed(500)
Cornell.boot <- bootpls(plsR(yCornell,XCornell,3), sim="permutation", stype="i", R=1000)
boot::boot.array(Cornell.boot, indices=TRUE)


# Graph of bootstrap distributions
boxplot(as.vector(Cornell.boot$t[,-1])~factor(rep(1:7,rep(1000,7))),main="Bootstrap distributions of standardised bj (j = 1, ..., 7).")
points(c(1:7),Cornell.boot$t0[-1],col="red",pch=19)
# Using the boxplots.bootpls function
boxplots.bootpls(Cornell.boot,indices=2:8)




library(boot)
plot(Cornell.boot,index=2)

qqnorm(Cornell.boot$t[,2],ylim=c(-1,1))
abline(h=Cornell.boot$t0[2],lty=2)
(sum(abs(Cornell.boot$t[,2])>=abs(Cornell.boot$t0[2]))+1)/(length(Cornell.boot$t[,2])+1)

rm(Cornell.boot)