A software tool for regression analysis and its assumptions

Abstract

Nowadays, among the forecasting methods, the most important one is the regression analysis. In this method, the aim is to estimate the population regression model as much as accurate by taking as basis the sample regression function. Its results are valid under certain assumptions and the violations of these assumptions cause the invalidity of some properties of the estimators. In this study, a new object-oriented program concentrated only on the regression analysis and its assumptions has been developed using Java, to carry out this analysis more easily and in a shorter time. In this program, regression model selection, regression and correlation analysis with Least Square method, one test for every assumption and solution methods has been presented. All the results of the analysis are illustrated by using a multiple regression example. © 2006 Asian Network for Scientific Information.