Parameter estimation of MO-qGEVL distribution under type II progressive censoring with application in environmental data

Abstract

Introducing new distributions is crucial for achieving more accurate fits to real-world data. In this context, many researchers have modified existing distributions by incorporating additional parameters. So a new distribution-called the Marshall-Olkin generalized extreme value distribution under linear normalization-is proposed and its parameters are estimated under type II progressive censoring by maximum likelihood estimation (MLE) and Bayesian estimation. However, the MLE of the original GEVL distribution cannot be obtained using classical numerical methods; so, the Whale optimization algorithm is utilized instead. Since humidity plays a vital role in the spread of forest fires, a real data set representing humidity levels in Los Angeles is investigated, showing the superiority of the proposed distribution compared to existing models. Moreover, both distribution fitting (return levels) and time series analysis are employed to forecast future humidity levels.