Dear community, 

in the lasso regression, which parameter should I vary to change how many variables are shrunken to zero?

I did not find a source saying that alpha is lambda in SPSS. Do you know it and maybe even have a source?

This is the code that I was using for now to choose lambda via cross validation: 

LINEAR_LASSO y WITH x1 x2 x3 
  /MODE CROSSVALID
  /ALPHA VALUES=0.01 TO 1 BY 0.01 METRIC=LINEAR
  /CRITERIA INTERCEPT=TRUE STANDARDIZE=TRUE TIMER=5.0 NFOLDS=5 STATE=0
  /PARTITION TRAINING=80 HOLDOUT=20.0
  /PRINT COMPARE
  /PLOT MSE R2.

Other questions I have (if someone happens to know a lot about LASSO):

• can I use this method for feature selection and then calculate a multiple linear regression or a logistic regression with the selected features?
• How many predictors would a regression have if LASSO really makes sense to use?

Thanks a a lot 

------------------------------
Anna Büssow
------------------------------