# [R] 2-dimensional probability density surface using thin-plate spline

impromptu@nifty.com impromptu at nifty.com
Tue Jan 21 10:49:03 CET 2003

```R-users
E-mail: r-help at stat.math.ethz.ch

I have a problem with usage of "gss". I would like to obtain
2-dimensional probability density surface using thin-plate spline.
My object is below:

function()
{
# (1)
library(gss)
set.seed(2523)
xx1 <- c(rnorm(60, mean=3, sd=2), rnorm(40 , mean=9, sd=1))
xx2 <- c(rnorm(60, mean=4, sd=2), rnorm(40 , mean=10, sd=1))
# (2)
br1 <- seq(from = floor(min(xx1)*2)/2, to = ceiling(max(xx1)*2)/2+0.5, by = 0.5)
br2 <- seq(from = floor(min(xx2)*2)/2, to = ceiling(max(xx2)*2)/2+0.5, by = 0.5)
data1 <- data.frame(x1 = xx1, x2 = xx2)
domain=data.frame( x1=c(br1, br1[length(br1)] ), x2=c(br2, br2[length(br2)] )) )
# (3)
midp1 <- br1[1:(length(br1)-1)]+(br1-br1)*0.5
midp2 <- br2[1:(length(br2)-1)]+(br2-br2)*0.5
midpg_expand.grid(midp1, midp2)
data2 <- data.frame(x1=midpg[,1], x2=midpg[,2])
ey <- dssden(fit1, data2)
ey_matrix(ey, ncol=length(midp2))
print(ey)
}

The result is:

Error in array(x, c(length(x), 1), if (!is.null(names(x))) list(names(x),  :
attempt to set an attribute on NULL

I am just wondering what causes this.

*****    Kunio Takezawa, Ph.D. (takezawa at affrc.go.jp)    *****