1
$\begingroup$

I am looking for a Least Angle Regression (LAR) packages in R or MATLAB which can be used for classification problems.

The only package that I currently know which fits this description is glmpath. The issue with this package is that it is a little old and somewhat limited in its scope (I am forced to rely on logistic regression for classification problems model).

I am wondering if anyone knows of other packages that allow me to run LAR on different types of classification models, such as Support Vector Machines (see The Entire Regularization Path for the Support Vector Machine).

The ideal package would allow me to run LAR-type algorithms for different types of classification models and also provide a function that can produce the full regularization path.

||cite|improve this question||||
$\endgroup$
  • $\begingroup$ I just saw that The Entire Regularization Path for the Support Vector Machine paper also refers to an R package named 'svmpath'. This answers some of my question, but any other packages that you guys know about would be very helpful. $\endgroup$ – Berk U. Jan 27 '13 at 23:33
  • 1
    $\begingroup$ GLMNET is another R package $\endgroup$ – probabilityislogic Jan 27 '13 at 23:39
  • $\begingroup$ The package LARS in R. I wondered why it wouldn't turn up on a google search of say 'least angle regression in R'... but for me it's the top result. Did you not google? $\endgroup$ – Glen_b -Reinstate Monica Jan 27 '13 at 23:46
  • $\begingroup$ @Glen_b I did, and it did turn up, but as far as I know the LARS package only handles regression problems and not classification problems - right? $\endgroup$ – Berk U. Jan 27 '13 at 23:56
  • $\begingroup$ My apologies. Well of course regression can be used for classification, but that's not what you mean. $\endgroup$ – Glen_b -Reinstate Monica Jan 28 '13 at 2:14
2
$\begingroup$

For those that might run into this, there is the SpaSM toolbox for matlab. It includes LAR and other methods of relevance. Otherwise there is also the Glmnet for matlab and the R version, they use coordinate descent (at least in the matlab version).

||cite|improve this answer||||
$\endgroup$

Your Answer

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

Not the answer you're looking for?Browse other questions tagged or ask your own question.