I am using the package rpart to explore various classification structures.
The call looks like:
seekhi1<-rpart(pvol~spec+a1+psize+eppres+numpt+icds+bivalcrt+stents+ppshare+
nhosp+nyrs,data=dat,method="class",
control=rpart.control(minsplit=30,xval=10))
The output is
1) root 198 87 1 (0.5606061 0.4393939)
2) psize=1,2 122 43 1 (0.6475410 0.3524590)
4) a1=3 23 5 1 (0.7826087 0.2173913) *
5) a1=1,2 99 38 1 (0.6161616 0.3838384)
10) numpt=1 53 16 1 (0.6981132 0.3018868) *
11) numpt=2 46 22 1 (0.5217391 0.4782609)
22) nhosp=1 18 5 1 (0.7222222 0.2777778) *
23) nhosp=2 28 11 2 (0.3928571 0.6071429) *
3) psize=3 76 32 2 (0.4210526 0.5789474)
6) spec=3 16 5 1 (0.6875000 0.3125000) *
7) spec=1,2 60 21 2 (0.3500000 0.6500000)
14) bivalcrt=2 34 16 2 (0.4705882 0.5294118)
28) stents=2 17 6 1 (0.6470588 0.3529412) *
29) stents=1 17 5 2 (0.2941176 0.7058824) *
15) bivalcrt=1 26 5 2 (0.1923077 0.8076923) *
I'd like to force rpart to use a specific variable, namely "spec"
as the
first split and then allow it to proceed as usual. Is there a way to do
this?
Thanks,
Bruce
[[alternative HTML version deleted]]