A Comparative Study of Estimation Methods for Parameters in Multiple Linear Regression Model

Abstract

Cankaya, S., Kayaalp, G.T., Sangun, L., Tahtali, Y. and Akar, M. 2006. A comparative study of estimation methods for parameters in multiple linear regression model. J. Appl. Anim. Res., 29: 43–47.

This paper investigated least squares method, non-parametric method and robust regression methods to predict the parameters of multiple regression models. To evaluate these methods, measurements of body weight, total length and fork length of fishes collected from Serranus cabrilla were used. In these regression models, body weight was dependent variable whereas total length and fork length were independent variables. The results show that non-parametric regression method, general additive model, has minimum R2 value and least median squares has maximum R² value, 0.334 and 0.855, respectively.

Keywords:

• Least squares
• least median squares
• least trimmed squares
• robust regression
• non-parametric multiple linear regression
• Serranus cabrilla