PPR的收敛性和全面攻击导弹数据处理

TIAN ZHENG EONG HAI WU (Northwestern Polytechnicai University, Xian, 710072)

本文证明了投影寻踪回归(Projection Pursuit Regression简称PPR)中岭函数为多项式形式时,PPR的L_2收敛性;对岭函数为多项式形式的投影寻踪回归给出了一种新的算法;应用这种算法对全向攻击导弹数据进行了处理,所获得的回归模型可供科研单位实际使用.

This paper proves the strong convergence of projection pursuit regression whose ridge functions are polynomials, and it establishes a new algorithm to projection persuit regression whose ridge functions are polynomials. As compared with the method established by Friedman and Stuntzle (1981), this algorithm features higher rate of convergence and easier for programming. At the end of this paper, the models of R_(max) and R_(min) are established.