Authors : U.N. Bassey and E.O. Effanga
Abstract: In this study, we reformulate the linear absolute regression problem as a linear goal-programming problem, whose associated dual problem is obtained and exploited computationally to estimate the regression parameters. It is shown that the goal programming method of estimating the regression parameters is better, in terms of the total absolute error, than the least squares method.
U.N. Bassey and E.O. Effanga , 2008. A Linear Goal Programming Model for the Linear Absolute Value Regression Problem. Journal of Modern Mathematics and Statistics, 2: 123-125.