頁籤選單縮合
題 名 | 改良式雙重平滑估計量=An Improved Version of the Double Smoothing Estimator |
---|---|
作 者 | 甘美蓮; | 書刊名 | 中國統計學報 |
卷 期 | 37:4 1999.12[民88.12] |
頁 次 | 頁385-400 |
分類號 | 319.51 |
關鍵詞 | 有限樣本性質; 無母數迴歸分析; 區域線性估計量; 漸近性質; 雙重平滑估計量; 邊界效應; Asymptotic properties; Boundary effects; Double smoothing estimator; Finite sample properties; Local linear estimator; Nonparametric regression; |
語 文 | 中文(Chinese) |
中文摘要 | 針對無母數迴歸函數估計,本文將Wu與Chu的雙重平滑估計方法加以改良,並證明其具有下列優點:有限樣本條件變異數與條件偏差量均為有界,與區域線性估計量有相同的漸近變異數及漸近偏差量,此外,其實務上產生的迴歸函數估計曲線比較平滑。摸擬結果顯示本文所提出的估計量較文獻中常見的兩種估計量有更小的樣本積分均方差。 |
英文摘要 | In the present paper we propose an improved version of Wu and Chu's double smoothing estimator for nonparametric regression functions. Our estimator has the following advantages. First, its finite sample conditional variance and bias are bounded. Second, it has the same asymptotic variance and bias as the local linear estimator. In addition, it usually produces smooth estimated regression function curves. Simulations demonstrate that the proposed improved estimator performs better than the other two existing approaches suggested to correct for the drawback of the local linear estimator that the conditional variance may become infinite in the finite sample case. |
本系統中英文摘要資訊取自各篇刊載內容。