International Journal of Science, Engineering and TechnologyAuthors: Chukwudi Anderson Ugomma, Samuel Chimuanya Chijioke
Abstract: This study presents a comparative analysis of the Inverse Weibull and Inverse Lognormal distributions using both simulated and real-world data with Maximum Likelihood Estimation (MLE). Model selection criteria included Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Anderson-Darling (AD), and Kolmogorov-Smirnov (KS) tests. Six simulated sample sizes (50, 100, 150, 200, 250, and 500) were used, with 1000 replications each. Results showed the Inverse Lognormal distribution consistently had lower AIC and BIC values at small to moderate sample sizes. Furthermore, real-world stock price data (sample size 100) from the Nigerian Stock Exchange was analyzed. Descriptive statistics and goodness-of-fit tests favored the Inverse Lognormal model. These findings support the utility of AIC, BIC, AD, and KS in lifetime data modeling and model selection.
DOI: https://doi.org/10.5281/zenodo.17627972
