Abstract
It is known that feature selection/screening for high-dimensional nonparametric models is an important but very difficult issue. In this paper, we first point out the limitations of existing screening methods. In particular, model-free sure independence screening methods, which are defined on random predictors, may completely miss some important features in the underlying nonparametric function when the predictors follow certain distributions. To overcome these limitations, we propose an ensemble screening procedure for nonparametric models. It elaborately combines several existing screening methods and outputs a result close to the best one of these methods. Numerical examples indicate that the proposed method is very competitive and has satisfactory performance even when existing methods fail.
Funder
National Key R&D Program of China
National Natural Science Foundation of China
Subject
General Mathematics,Engineering (miscellaneous),Computer Science (miscellaneous)
Reference32 articles.
1. Santner, T.J., Williams, B.J., and Notz, W.I. (2018). The Design and Analysis of Computer Experiments, Springer. [2nd ed.].
2. Cressie, N.A.C. (2015). Statistics for Spatial Data, John Wiley & Sons. [Revised ed.].
3. Friedman, J., Hastie, T., and Tibshirani, R. (2008). The Elements of Statistical Learning: Data Mining, Inference and Prediction, Springer. [2nd ed.].
4. Sure independence screening for ultrahigh dimensional feature space;Fan;J. R. Stat. Soc. Ser. B,2008
5. Sure independence screening in generalized linear models with NP-dimensionality;Fan;Ann. Stat.,2010
Cited by
1 articles.
订阅此论文施引文献
订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献