![IconGems-Print](images/IconGems-Print.png)
30 – Bayesian Modeling and Time Series
A Least Deviation Estimation Approach for Multiple Regression
Silvey Shamsi
Center for Business and Economic Research (CBER), Ball State University
Mian Arif Shams Adnan
Department of Computer Science, Ball State University
Rahmatullah Imon
Department of Mathematical Sciences, Ball State University
Regression estimates of the Multiple Regression suffer from the exaggeration of the contribution of the extreme observation(s) as well as the outlier(s). Unlike the traditional fitting procedure of multiple regression based on the least squares estimates, it uses least deviation method for minimizing the total sum of errors. It also suggests some new measures of Coefficient of Determination.