I don't know how many steps are necessary for lars() to select the variables till the algorithm proceeds to the saturated fit (especially using the lars package in R)? Can anyone give me a hint?
$\begingroup$
$\endgroup$
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
1
The second sentence in the second section of The LARS paper says that the LARS algorithm requires $m$ steps, where $m$ is the number of covariates.
answered Dec 19, 2012 at 18:34
-
$\begingroup$ Thanks for reply.I have seen that when the number of predictors(m) is much larger than the number of observations(n) in high dimensional data then lars gets at most n steps, but I am not sure how I can define the exact number of steps in order to have the saturated fit. Also is it possible to give me a hint how I can define the parameter lambda for lasso in order to have the saturated fit?(any document that explain it). thanks in advance $\endgroup$– user22Commented Dec 19, 2012 at 19:12