How to fit a linear model including the expectation and a transformation

Question

I have to fit a linear model for

E(ᴪ(Y)) = β0+ β1X1+ β2X2+ β3X3

where ᴪ is a modified Box-Cox transformation given by:

and β0,1,2,3 are parameters. My data frame is a shown as below which I read into R as

data<-read.table("surveydata.txt", header=TRUE)

I'm fairly new to R and I haven't been able to fit the model. I know how to fit a model using using lm, but wasn't sure how I'll fit it while including the transformation and the expectation.

So far I've tried:

>lm(mean(boxcox(Y))~X1+X2+X3,data)
Error: $ operator is invalid for atomic vectors

Welcome to stackoverflow! Please see http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example for suggestions on how people can reproduce your problem. A good first step would be including the dataset with `dput()`, rather than as an image. — Andy Clifton, Apr 07 '14 at 18:01
Why don't you just run `boxcox(Y ~ X1 + X1 + X3, data, plotit = F)` ? — David Arenburg, Apr 07 '14 at 18:33
David, I did that and it prints the lambda vs log likelihood. How do i get from this to a fitted model? — user3506632, Apr 07 '14 at 18:41
You obviosuly didn't do that with `plotit = F` like i showed. If you do that that way, you"ll get your fitted model. Run `boxcox(Y ~ X1 + X1 + X3, data, plotit = F)$y` — David Arenburg, Apr 07 '14 at 19:34

score 1 · Answer 1 · answered Apr 07 '14 at 20:29

1

boxcox(Y ~ X1 + X1 + X3, data, plotit = F)$y

answered Apr 07 '14 at 20:29

David Arenburg

91,361
17
137
196

If this the answer you were looking for, please select it as one, thanks – David Arenburg Apr 08 '14 at 07:18

How to fit a linear model including the expectation and a transformation

1 Answers1