Modified outlier diagnostics for the extended exponential regression model with interval and right-censored data

Abstract

. This research will explore the modified outlier diagnostics for the exponentiated exponential regression (EER) model with interval and right-censored data using adjusted residuals and some influence diagnostic measures. The EER model accommodates a monotonically increasing and decreasing hazard rate and reduces to the exponential as a special case. The proposed diagnostics utilize the adjusted residuals based on the bias-corrected bootstrap harmonic mean with random imputation and the measure of curvature. A simulation study is conducted to assess the performances of different methods at various sample sizes, contamination levels, and censoring proportions. The results indicate that the adjusted Cox-Snell residual, martingale residual, and d ma x work best to detect outliers in a data set. Finally, the proposed methods are applied to a data set from the Diabetic Retinopathy (DRS) study.