A COMPARATIVE ANALYSIS ON LIFETIME DISTRIBUTIONS
Faculty
Department
Year of Publication
Publication Type
Abstract
This study investigates the application of Exponential, Weibull, and Gamma distributions in modeling lifetime data. The primary objective is to compare these distributions in real world survival analysis and reliability modeling. The study utilizes secondary data from published research, including survival times of head and neck cancer patients and waiting times of bank customers .
Maximum Likelihood Estimation (MLE) was employed to estimate distribution parameters, and model comparisons were performed using Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) to determine the best fitting distribution.
Findings reveal that while the Exponential distribution provides a simple model for constant failure rates, the Weibull distribution offers greater flexibility in modeling varying failure rates. The Gamma distribution demonstrates robust applicability in complex survival data. Results indicate that the Weibull and Gamma distributions provide superior fits in most real world cases.
This study contributes to the field of survival analysis and reliability engineering by providing insights into selecting appropriate lifetime distributions for different applications. The findings have practical implications for fields such as healthcare, engineering, and risk assessment, where accurate lifetime modeling is crucial for decision making.
Maximum Likelihood Estimation (MLE) was employed to estimate distribution parameters, and model comparisons were performed using Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) to determine the best fitting distribution.
Findings reveal that while the Exponential distribution provides a simple model for constant failure rates, the Weibull distribution offers greater flexibility in modeling varying failure rates. The Gamma distribution demonstrates robust applicability in complex survival data. Results indicate that the Weibull and Gamma distributions provide superior fits in most real world cases.
This study contributes to the field of survival analysis and reliability engineering by providing insights into selecting appropriate lifetime distributions for different applications. The findings have practical implications for fields such as healthcare, engineering, and risk assessment, where accurate lifetime modeling is crucial for decision making.
Supervisor(s)
co-supervisor