|
In this paper,we combine LAD-LASSO estimation and nonnegative constraint estimation,to propose a robust estimation which can do parameter estimation and variable selection in non negative problem.Compared with LAD-LASSO, he can better handle some non negative problems in economy. And compared with non negative estimation,it can do variable selection.With easily estimated tuning parameters,the non negative enjoys oracle property.Furthermore,we propose a non negative coordinate descent algorithm and do some data simulation. We also applied the model to stock index tracking and compared with non negative LASSO. |
|
Keywords:Application statistics, Nonnegative LADLASSO, Robust estimation, Index tracking, Variable selection |
|