Estimation Of The Linear EV Regression Model With Censored Data

For a long time, the linear EV regression model has received extensive attention by many researchers due to its simple form and a wide range of application. Partic-ularly, it has been widely used in life analysis, reliability analysis and so on. Thus, many scholars at home and abroad have put forward the different estimation methods for the linear EV regression model. However, it is a problem to estimate and forecast the time of certain event that occurs in many fields, and the response variable is often censored or truncated, which leads that the observed data is incomplete. Under some cases, one of the most important cases is that the variable is right censored. For these censored data studies, research scholars have proposed many kinds of models. So it is necessary for us to do some inferences under the linear EV regression model with censored data.Considering the characteristic of linear EV regression model with censored data, in this paper, one special case is considered, that is, when the variance of the mea-surement error of the linear EV regression model with censored data is unknown, the method of observing variables repeatedly is used to estimate it. Thereafter, we derive the moment estimation and the modified least squared estimation for the parameters respectively, so we can eliminate the assumption that the variance of the measurement error is known in some literatures. And under some regularity conditions, we prove the asymptotic properties of the parameter estimation. At last, we illustrate that the method proposed in this paper has good validity and feasibility through data to simu-late the moment estimation and the modified least squared estimation.