please help me
again I have a problem.
I want to do cross-validation for multinomial log. reg. I have a response
variabe with 7 levels (Z((a,b,c,d,e,f,g)) and 4 predictor(1 classifier and 3
continuous). I did:
#data with 100 observations
> library('bootstrap')
>x<-matrix(c(data$H, data$K, data$P, data$W),100,4)
>y<-data$Z
>theta.fit <- function(x,y){lsfit(x,y)}
>theta.predict <- function(fit,x){cbind(1,x)%*%fit$coef}
>sq.err <- function(y,yhat) { (y-yhat)^2}
>results <- bootpred(x,y,50,theta.fit,theta.predict, sq.err)
>miss.clas <- function(y,yhat){ 1*(yhat!=y)}
>results <- bootpred(x,y,50,theta.fit,theta.predict, miss.clas)
do this form correct for my data?
if yes;
but why after both results, R give me error?
Error in lsfit(x, y) : NA/NaN/Inf in foreign function call (arg 1)
In addition: Warning messages:
1: In storage.mode(x) <- "double" : NAs introduced by coercion
2: In storage.mode(y) <- "double" : NAs introduced by coercion
Thanks alot
Azam
[[alternative HTML version deleted]]