Modeling the Mixture of Lognormal, Gamma and Weibull Distributions for System Reliability Performance

Authors

  • S. I. Kalu School of Science & Industrial Technology, Department of Statistics, Ogbonnaya Onu Polytechnic, Aba Author
  • U. E. Amaechi School of Science & Industrial Technology, Department of Statistics, Ogbonnaya Onu Polytechnics, Aba Author

DOI:

https://doi.org/10.5281/zenodo.17605524

Keywords:

System reliability, survival analysis, UCI Machine Learning Repository, mixture distribution, mixture model, Weibull, Lognormal, Gamma distribution

Abstract

This research develops a statistical framework for modeling system reliability using a mixture of Lognormal, Gamma, and Weibull distributions. The goal is to improve reliability estimation for complex systems exhibiting multiple failure modes. Data for the analysis is obtained from the UCI Machine Learning Repository, specifically the 'Simulated Data for Survival Modeling' dataset. Each distribution is fitted separately, and their mixture is then estimated using maximum likelihood methods. Comparative analyses are performed based on goodness-of-fit indices (AIC, BIC, log-likelihood) and reliability metrics such as mean time to failure (MTTF). Results demonstrate that the mixture model provides superior fit and predictive reliability over single distribution models.

Downloads

Published

2025-11-14

How to Cite

Kalu, S. I., & Amaechi, U. E. (2025). Modeling the Mixture of Lognormal, Gamma and Weibull Distributions for System Reliability Performance. International Journal of Innovative Engineering Research and Development, 1(2), 1-19. https://doi.org/10.5281/zenodo.17605524