L1 regularization reduces the magnitude of weights but does not make them zero. Is there a way we can make them zero. Will it be better than normal L1 regularization

  • $\begingroup$ Increase the penalty, regularization parameter enough and all the coefficients will be zero. $\endgroup$ – Matthew Gunn Aug 16 '17 at 17:55
  • $\begingroup$ @MatthewGunn What is the maximum value of parameter. Is it between 0 and 1? $\endgroup$ – shaifali Gupta Aug 16 '17 at 18:06
  • $\begingroup$ I assume you're referring to LASSO regression? If so, there's no upper bound on the regularization parameter. $\endgroup$ – Matthew Gunn Aug 16 '17 at 18:33
  • $\begingroup$ @MatthewGunn Then how to choose the value of regularization parameter. because there is no limit on the number of cases that can be tried $\endgroup$ – shaifali Gupta Aug 18 '17 at 14:37
  • 1
    $\begingroup$ @Ben Looking back, I shouldn't have written that. In practice, it often looks convex (or quasiconvex) but there's no requirement. $\endgroup$ – Matthew Gunn Sep 5 '17 at 18:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.