Showing 1–1 of 1 results for author: Nassiri, V

An efficient algorithm for structured sparse quantile regression

Authors: Vahid Nassiri, Ignace Loris

Abstract: Quantile regression is studied in combination with a penalty which promotes structured (or group) sparsity. A mixed $\ell_{1,\infty}$-norm on the parameter vector is used to impose structured sparsity on the traditional quantile regression problem. An algorithm is derived to calculate the piece-wise linear solution path of the corresponding minimization problem. A Matlab implementation of the prop… ▽ More Quantile regression is studied in combination with a penalty which promotes structured (or group) sparsity. A mixed $\ell_{1,\infty}$-norm on the parameter vector is used to impose structured sparsity on the traditional quantile regression problem. An algorithm is derived to calculate the piece-wise linear solution path of the corresponding minimization problem. A Matlab implementation of the proposed algorithm is provided and some applications of the methods are also studied. △ Less

Submitted 25 February, 2013; originally announced February 2013.

Comments: 15 pages, 4 figures