The existing methods for tting mixture regression models assume a normal dis- tribution for error and then estimate the regression parameters by the maximum likelihood estimate (MLE). In this article, we demonstrate ...