Bezostaja buğdayının fiyatına etki eden bazı donelerin alternatif regresyon yöntemleriyle karşılaştırılması

Abstract

Regression analysis is a technique which examines the reason and result between two or more variables and creates a mathematical model in order to make a prediction on the subject. In this study, the linear relationship between the variables that affect the price and the price itself has been examined. Multiple regression analysis modeling for 12 months Bezostaja wheat varieties data and parameter estimation methods were tried to be compared. Estimators of model parameters; Least squares method (LS), Smallest absolute deviation regression method (LAD), non-parametric regression method, Mregression method and Bayes regression method. In the multiple linear regression model, estimates were made by applying these methods to the available data, and the appropriate method was proposed by comparing the results. Under these circumstances, although non-parametric regression estimation method seems to be a convenient estimator, according to the MSE and AIC criteria, LS estimation method steps forward as the most convenient method for this data set.