Programming the statistical procedures from SAS

calculate the Generalised Cross validation (GSV) in proc reg to obtain the value of k

Reply
New Contributor kwa
New Contributor
Posts: 4

calculate the Generalised Cross validation (GSV) in proc reg to obtain the value of k

Dear community members I am using PROC REG to perform a ridge regression. I was wondering what command to use to calculate the Generalised Cross validation (GSV) in proc reg to obtain the value of k that yields the smallest GCV criterion.

I know how do so in R e.g in R we would

1. Create a regression object using the lm.ridge() function. For example:
fox_ridge<-lm.ridge((bGSH~ln_bInAs + ln_bMMA + ln_bDMA + sex + cig + ln_age, lambda=seq(5,100,1))
##Note that I have specified a range of values for k (called “lambda” in R). GCV tends to select values of k close to 0, so it is best to restrict the possible range of k values.

2. Obtain GCV criterion for each value of k using the code $GCV following your regression object. For example:
fox_ridge$GCV

3. Select the value of k that yields the smallest GCV criterion


I however would like to know how to do so in SAS PROC reg command
Ask a Question
Discussion stats
  • 0 replies
  • 121 views
  • 0 likes
  • 1 in conversation