Cox proportional hazard (PH) model is the most popular model for modeling survival data. Accelerated failure time (AFT) model basically used for analyzing failure time data in industrial equipments may be an alternative to Cox PH model. In this paper, an attempt has been made to compare the Cox PH model and different forms of AFT models (exponential, Weibull, log-logistic, log-normal, gamma, etc.) in the context of survival data of esophagus cancer patient of Assam. Akaike information criterion (AIC), Bayesian information criterion (BIC) and Cox-Snell residuals are used to identify the best fitted model. The results of the present study also establish the fact that the Cox PH model is the best fitted model for analyzing survival data of esophagus cancer patients.

Cox PH model, accelerated failure time model, AIC, BIC, esophagus cancer patients.