Influence diagnostics in Log-Logistic regression model with censored data

Log-Logistic regression model arise in several areas of application. Traditional estimation methods for Log-Logistic regression model are sensitive to influential observations. Such bizarre observations can isolate analysis and lead to incorrect conclusions and actions. We suggest local influence diagnostics for identifying unusual observations in Log-Logistic regression model with censored data. The diagnostic methods under the perturbation scheme of case weight, explanatory and response variables are derived. Computational statistical measures are proposed that make the procedures practicable. Moreover, Generalized Cook's distance and One-step Newton-Raphson method are also studied. Finally, a real data set and simulation study is presented. The results of illustrative example and simulation scheme clearly reveal that the proposed diagnostic methods under normal curvature perform better than others. (C) 2021 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Alexandria University.

Automated summary

The study focused on diagnostic methods for identifying unusual observations in the Log-Logistic regression model, proposing a new approach based on local influence diagnostics, and demonstrating its superiority through an illustrative example and simulation study.