Variable Selection for a Nonparametric Nonlinear System by Directional Regression

The importance of discovering significant variables from a large candidate pool is now widely recognized in many fields. There exist a number of algorithms for variable selections in the literature. Some are computationally efficient but only provide a necessary condition, not a sufficient and necessary condition, for testing if a variable contributes or not. The others are computationally expense. The goal of the paper is to develop a directional variable selection algorithm that performs similar to or better than the leading algorithms for variable selections, but under weaker technical assumptions and with a much reduced computational complexity.