Inference for partially observed competing risks model for Kumaraswamy distribution under generalized progressive...

Area of Science:

• Statistics
• Reliability Engineering
• Survival Analysis

Background:

• Competing risks models are crucial for analyzing multiple failure causes.
• The Kumaraswamy distribution offers flexibility for modeling lifetime data.
• Generalized progressive hybrid censoring is an efficient data collection scheme.

Purpose of the Study:

• To develop statistical inference methods for a competing risks model with Kumaraswamy latent failure times.
• To address challenges posed by partially observed failure causes and censoring.
• To investigate the impact of order-restricted shape parameters on inference.

Main Methods:

• Establishing the existence and uniqueness of Maximum Likelihood Estimators (MLEs).
• Utilizing asymptotic distribution theory for confidence interval construction.
• Computing Bayesian estimators and credible intervals.
• Employing Monte Carlo simulations to assess estimator performance.

Main Results:

• The existence and uniqueness of MLEs for model parameters are proven.
• Asymptotic confidence intervals are derived.
• Bayesian estimation provides credible intervals for parameters.
• Simulations demonstrate the performance of various estimation techniques.

Conclusions:

• The proposed methods provide valid statistical inference for the studied competing risks model.
• The findings are illustrated with a real-world data analysis.
• The study contributes to the statistical analysis of complex failure data.