89 lines
3.4 KiB
Plaintext
89 lines
3.4 KiB
Plaintext
|
|
R Under development (unstable) (2019-04-05 r76323) -- "Unsuffered Consequences"
|
|
Copyright (C) 2019 The R Foundation for Statistical Computing
|
|
Platform: x86_64-pc-linux-gnu (64-bit)
|
|
|
|
R is free software and comes with ABSOLUTELY NO WARRANTY.
|
|
You are welcome to redistribute it under certain conditions.
|
|
Type 'license()' or 'licence()' for distribution details.
|
|
|
|
R is a collaborative project with many contributors.
|
|
Type 'contributors()' for more information and
|
|
'citation()' on how to cite R or R packages in publications.
|
|
|
|
Type 'demo()' for some demos, 'help()' for on-line help, or
|
|
'help.start()' for an HTML browser interface to help.
|
|
Type 'q()' to quit R.
|
|
|
|
> #
|
|
> # Test weights in a regression problem
|
|
> #
|
|
> library(rpart)
|
|
> set.seed(10)
|
|
>
|
|
> mystate <- data.frame(state.x77, region=factor(state.region))
|
|
> names(mystate) <- c("population","income" , "illiteracy","life" ,
|
|
+ "murder", "hs.grad", "frost", "area", "region")
|
|
>
|
|
> xgrp <- rep(1:10,5)
|
|
> fit4 <- rpart(income ~ population + region + illiteracy +life + murder +
|
|
+ hs.grad + frost , mystate,
|
|
+ control=rpart.control(minsplit=10, xval=xgrp))
|
|
> wts <- rep(3, nrow(mystate))
|
|
> fit4b <- rpart(income ~ population + region + illiteracy +life + murder +
|
|
+ hs.grad + frost , mystate,
|
|
+ control=rpart.control(minsplit=10, xval=xgrp), weights=wts)
|
|
> fit4b$frame$wt <- fit4b$frame$wt/3
|
|
> fit4b$frame$dev <- fit4b$frame$dev/3
|
|
> fit4b$cptable[,5] <- fit4b$cptable[,5] * sqrt(3)
|
|
> temp <- c('frame', 'where', 'splits', 'csplit', 'cptable')
|
|
> all.equal(fit4[temp], fit4b[temp])
|
|
[1] TRUE
|
|
>
|
|
>
|
|
> # Next is a very simple case, but worth keeping
|
|
> dummy <- data.frame(y=1:10, x1=c(10:4, 1:3), x2=c(1,3,5,7,9,2,4,6,8,0))
|
|
>
|
|
> xx1 <- rpart(y ~ x1 + x2, dummy, minsplit=4, xval=0)
|
|
> xx2 <- rpart(y ~ x1 + x2, dummy, weights=rep(2,10), minsplit=4, xval=0)
|
|
>
|
|
> all.equal(xx1$frame$dev, c(82.5, 10, 2, .5, 10, .5, 2))
|
|
[1] TRUE
|
|
> all.equal(xx2$frame$dev, c(82.5, 10, 2, .5, 10, .5, 2)*2)
|
|
[1] TRUE
|
|
>
|
|
> # Now for a set of non-equal weights
|
|
> # We need to set maxcompete=3 because there just happens to be, in one
|
|
> # of the lower nodes, an exact tie between variables "life" and "murder".
|
|
> # Round off error causes fit5 to choose one and fit5b the other.
|
|
> # Later -- cut it back to maxdepth=3 for the same reason (a tie).
|
|
> #
|
|
> nn <- nrow(mystate)
|
|
> wts <- rep(1:5, length=nn)
|
|
> temp <- rep(1:nn, wts) #row replicates
|
|
> xgrp <- rep(1:10, length=nn)
|
|
> xgrp2<- rep(xgrp, wts)
|
|
> tempc <- rpart.control(minsplit=2, xval=xgrp2, maxsurrogate=0,
|
|
+ maxcompete=3, maxdepth=3)
|
|
> # Direct: replicate rows in the data set, and use unweighted
|
|
> fit5 <- rpart(income ~ population + region + illiteracy +life + murder +
|
|
+ hs.grad + frost , data=mystate[temp,], control=tempc)
|
|
> # Weighted
|
|
> tempc <- rpart.control(minsplit=2, xval=xgrp, maxsurrogate=0,
|
|
+ maxcompete=3, maxdepth=3)
|
|
> fit5b <- rpart(income ~ population + region + illiteracy +life + murder +
|
|
+ hs.grad + frost , data=mystate, control=tempc,
|
|
+ weights=wts)
|
|
> all.equal(fit5$frame[-2], fit5b$frame[-2]) # the "n" component won't match
|
|
[1] TRUE
|
|
> all.equal(fit5$cptable, fit5b$cptable)
|
|
[1] TRUE
|
|
> all.equal(fit5$splits[,-1],fit5b$splits[,-1])
|
|
[1] TRUE
|
|
> all.equal(fit5$csplit, fit5b$csplit)
|
|
[1] TRUE
|
|
>
|
|
> proc.time()
|
|
user system elapsed
|
|
0.102 0.031 0.128
|