Can MARS be improved with B-splines?

摘要

Et y represent a signle response variable that depends on a vector of n predictor variables x chemical bounds (x_1,... , x_n). Assume we are given M data poinmts x_m chemical bounds (x_(1m),..., x_(nm)), m chemical bounds 1, ... , Min the predictor domain and response values y_m chemical bounds f (x_m) + #epsilon# _m, where f is unknown but assumed smooth in E~((n)) (euclidian n space) and {#epslion# _m) are mean zero random variables. The goal is to determine, a which is a reasonable approximation of f over the predictor domain.