การเปรียบเทียบการคัดเลือกตัวแบบการถดถอยแบบริดจ์ โดยใช้เกณฑ์ซีพี และสถิติทดสอบซีพี
คำสำคัญ:
Ridge regression, Cp criteria, Cp test statisticsบทคัดย่อ
The purpose of this study is to compare the independent variable selection in analyzing multiple linear Ridge Regression by Hoerl and Kennard, Hoerl Kennard and Baldwin, Lawless and Wang, Khalaf and Shukur and Ordinary Least Square methods (OLS). Model selection is conducted by forward selection and backward elimination methods using the Cp criteria and Cp test statistics. The simulation results can be concluded that in all sample sizes, the percentage of selecting the correct model based on Cp test statistics by the forward selection and backward elimination method was higher than basing on the Cp criteria. The underspecified model and the misspecified model are not found in this study, only found overspecified model which its effects are less serious than those of the underspecified models. In addition, it is found that, all correlations between and give similar correct percentage. Various methods in estimate ridge parameter have percentage of select the correct model higher than OLS when small sample size and no difference when the sample size is larger.
เอกสารอ้างอิง
2. Hoerl AE, Kennard RW, Baldwin K. Ridge Regression: Some Simulations. Communications in Statistics, 1975; 4(2): 105-123.
3. Lawless J, Wang PA. Simulation study of ridge and other regression Estimators. Communication Statistics – Theory and Methods, 1976; 5(4): 307-323.
4. Khalaf G, Shukur G. Choosing ridge parameter for regression Problem. Communication Statistics in Theory and Methods, 2005; 34: 1177-1182.
5. Nisachon N. A Comparison of Variable Selection with Multicollinearity by Ridge Regression and Tabu Search [Msc thesis]. Bangkok: National Institute of Development Administration; 2013. Thai.
6. McQuarrie A, Robert S, Tsai CL. The model Selection Criterion AlCu. Statistics and Probability Letters, 1997; 34: 2585-292.
7. Sheather SJ. A Modern Approach to Regression with R. New York: Springer; 2009.
8. Mallows CL. Some Comments on CP. Technometrics, 1973; 15(4): 661-675
9. Songsiri T. Regression Analysis .3rd ed. Kasetsart University (BKK); 2005.
10. Rotjana K. A Comparison of the Model Selection Criterion in Multiple Linear Regression Analysis [Msc thesis]. Bangkok: Thammasart University; 2006. Thai.
11. Akaike H. Information Theory and an Extension of the Maximum Likelihood Principle. In 2nd International Symposium on Information Theory. B.N. Petrov and F.Cski, eds. Akademiai Kiado, Budapest; 1973; 15(4): 267-281
12. Palakorn S. Test Statistics for Selecting Multiple Linear Regression Models [Msc thesis]. Bangkok: National Institute of Development Administration; 2010. Thai.
