Comparison of some censored regression models with application

Abstract EN

Censored regression models are among the critical statistical models used in many studies and research, mainly in which the data are restricted in one part and not another.

Hence, it is difficult to use and apply traditional regression models to those data to analyze and study the relationship between the variables affecting them.

Many well-known censored regression models, such as the Tobit model, are the most widely used among these models.

In this paper, we used extended models of statistical distributions transformed into regression models capable of analyzing survival and reliability functions within the censored data.

They can be considered as censored regression models, including the Log-BXII Weibull model from the Weibull distribution and the LogBXIIEE model from the Exponentiated Exponential distribution in addition to the Tobit model and estimating the parameters of these models by the (MLE) method and applying these models to the data of (259) patients with renal failure (156 males and 103 females) of the dependent variable (Urea) and comparing the results of the three models through the use of comparison criteria (AIC, and H-QIC).

Using SAS software showed that the three models are equivalent and parallel in importance, meaning researchers can use any new model rather than the Tobit, the most widely used censored regression model.