Nonparametric estimation of the location of a maximum in a response surface

详细信息    查看全文

• 作者：Facer ; Matthew R. ; Mü ; ller ; Hans-Georg
• 关键词：62G08 ; 62G15 ; 62H12 ; 62K20 ; Nonparametric regression ; Maximal response ; Kernel method ; Local least squares ; Surface fitting ; AIDS incidence
• 刊名：Journal of Multivariate Analysis
• 出版年：2003
• 出版时间：October, 2003
• 年：2003
• 卷：87
• 期：1
• 页码：191-217
• 全文大小：587 K

文摘

We explore a nonparametric version of response surface analysis. Estimates for the location where maximum response occurs are proposed and their asymptotic distribution is investigated. The proposed estimates are based on kernel and local least squares methods. We construct asymptotic confidence regions for the location and include comparisons with the quadratic response surface approach. The methods are illustrated for the two-dimensional case with AIDS incidence data, where the point of maximum incidence is of interest.