鄭明燕2006-07-262018-06-282006-07-262018-06-282004-10-31http://ntur.lib.ntu.edu.tw//handle/246246/21001在生物、經濟等領域中，多維變量的值域邊界是一個重要問題，我們提出一種 計算簡易的估計方法並且證明它有最佳的漸近收斂速度。非參數曲面估計已被廣泛應用於 實務，其中多維區域線性迴歸估計有許多優點，我們提出數種降低多維區域線性迴歸估計 變異方法，發現皆可大幅降低漸近變異而不影響漸近誤差。In biological and econometrical sciences, it is often important to know the support boundary of a multivariate random vector. We propose a method that is easy to compute and prove that it achieves the optimal rate of convergence. Nonparametric estimation of surfaces has been widely used in practice. Multivariate local linear regression has many advantages. We suggest methods to reduce variance of multivariate local linear smoothers. It is shown that they all substantially enhances the stability while keeping the asymptotic bias unchanged.application/pdf29302 bytesapplication/pdfzh-TW國立臺灣大學數學系暨研究所多維區域線性迴歸非參數平滑方法值域邊界降變異法multivariate local linear regressionnonparametric smoothingsupport boundaryvariance reductionreporthttp://ntur.lib.ntu.edu.tw/bitstream/246246/21001/1/922118M002012.pdf