兩種脊迴歸模式與複線性迴歸模式之交互驗證比較

A Cross-Validation Comparison of Two Kinds of Ridge Regression and Multiple Regression

10.6773/JRMS.200912.0077

蔡顯麞(Hsien-Chang Tsai)；劉湘川(Hsiang-Chuan Liu)

複線性迴歸模式 ； 脊迴歸模式 ； 交互驗證 ； Multiple regression ； ridge regression ； cross-validation

測驗統計年刊

17期_下（2009 / 12 / 01）

77 - 83

繁體中文

本研究採用教育測驗資料，以十折交互驗證最小誤差平方法，比較兩種脊迴歸模式；新近被提出漸近最小迴歸係數離差均方脊迴歸模式（MS法），及眾所週知之HK脊迴歸模式（HK法）與傳統常用之複線性迴歸模式（MT法），實驗結果顯示MS脊迴歸模式有最佳之表現，並再度驗證HK脊迴歸模式優於傳統常用之複線性迴歸模式。

In this study, an educational testing data experiment by using a 10-fold cross-validation mean square error (MSE) is conducted. The performances of two ridge regression models, the new proposed MS ridge model and the well-known HK ridge model, and the traditional multiple linear regression model, MT model, are compared. Experimental result shows that the MS ridge model has the best performance, and the HK ridge model is better than the traditional multiple linear regression model, MT model.

1. Hoerl, A. E.,Kennard, R. W.(1970).Ridge Regression: Based estimation for nonorthogonal problems.Technometrics,12,55-67.
2. Hoerl, A. E.,Kennard, R. W.(1975).Ridge Regression: some simulations.Comm. In Statist,4,105-123.
3. Kohavi, R.(1995).A study of cross-validation and bootstrap for accuracy estimation and model selection.Proceedings of the Fourteenth International Joint Conference on Artificial Intelligence,San Mateo:
4. Theoblaldm, C. M.(1974).Generalization of mean square error applied to ridge regression.Journal of the Royal Statistical Society B,36,103-106.
5. 劉仁沛(1978)。台北市，國立臺灣大學農藝研究試驗統計組。
6. 劉湘川(1993)。脊迴歸分析蒙地卡羅模擬試驗之研究。測驗統計年刊，1，103-132。