An Alternative Method for Least-Squares Fitting of Parametric Survival Models

The problem of estimation of parameters of the two well-known three parameter models of force of mortality, viz., Weibull, and Makeham models, is considered in this paper. The method of least-squares, commonly used for estimation of parameters of these models, involves log-transformation of the force of mortality data. We propose an efficient numerical scheme for solving the problem. The proposed method involves fitting a series of linear regressions, and finding the least-squares estimates of all the parameters by the method of Golden Section search for computer minimization. Some examples of the proposed method will be given to compare the existing least-squares method with the proposed procedure.