2013/12/09 by Md Hasinur Rahaman Khan, Khan, Md Hasinur Rahaman, J.E. Shaw +1
Decision Sciences · Mathematics · #Advanced Statistical Methods and Models #Efficiency Analysis Using DEA #FOS: Computer and information sciences #Methodology (stat.ME) #Statistical Methods and Bayesian Inference #Statistical Methods and Inference
paper · pdf · doi:10.48550/arxiv.1312.2533
openalex publication_date 2013/12/09 · openalex created_date 2022/10/04 · openalex updated_date 2026/07/28
When observations are subject to right censoring, weighted least squares with\nappropriate weights (to adjust for censoring) is sometimes used for parameter\nestimation. With Stute's weighted least squares method, when the largest\nobservation is censored (Y(n)+), it is natural to apply the\nredistribution to the right algorithm of Efron (1967). However, Efron's\nredistribution algorithm can lead to bias and inefficiency in estimation. This\nstudy explains the issues clearly and proposes some alternative ways of\ntreating Y(n)+. The first four proposed approaches are based on the well\nknown Buckley--James (1979) method of imputation with the Efron's tail\ncorrection and the last approach is indirectly based on a general mean\nimputation technique in literature. All the new schemes use penalized weighted\nleast squares optimized by quadratic programming implemented with the\naccelerated failure time models. Furthermore, two novel additional imputation\napproaches are proposed to impute the tail tied censored observations that are\noften found in survival analysis with heavy censoring. Several simulation\nstudies and real data analysis demonstrated that the proposed approaches\ngenerally outperform Efron's redistribution approach and lead to considerably\nsmaller mean squared error and bias estimates.\n