Parametric Models and Future Event Prediction Base on Right Censored Data

摘要

In this paper, we developed a parametric displacement models base on the time that the Internally Displaced Persons (IDPs) took to return from IDPs Camps to their ancestral homes in Northern Uganda. The objective is to analyze the displaced proportion of the IDPs using suitable time-to-event parametric models. The accelerated failure time (AFT) models (Weibull, Exponential and log-logistic) were considered. A retrospective data of seven years study of 590 subjects is considered. Maximum likelihood method together with the Davidon-Fletcher- Powell optimization algorithm in MATLAB is used in the estimation of the parameters of the models. The estimated displaced proportions of these AFT models are used in predicting the displaced proportion of the IDPs at a time t. Weibull and exponential models provided better estimates of the displaced proportion of the IDPs due to their good convergence power to four decimal points and predicted the 2027 and 2044 respectively as the year when the displaced proportion can be approximated to be zero.