CHANGE POINT DETECTION USING SKEWED DISTRIBUTION IN THE FRAMEWORK OF MARSHALL – OLKIN GENERALIZED DISTRIBUTION

Abstract

{A novel change point detection framework based on the Marshall – Olkin Generalized Distribution integrates parametric and non-parametric methods, improving detection under skewed and heavy – tailed data.} 

This study proposes a novel framework for Change Point Detection (CPD) using the Marshall-Olkin Generalized Distribution (MOGD), a flexible distribution family that accommodates skewness and heavy tailed behaviour common features in real world data. Traditional CPD methods often rely on symmetric or normal distribution assumptions, which may fail to detect structural changes accurately in such contexts. The proposed approach integrates both parametric and non-parametric techniques, employing the Likelihood Ratio Test and Bayesian inference to identify change points in time series data modeled by MOGD. This dual method strategy enhances sensitivity to shifts in distributional behaviour, especially in datasets with asymmetry or extreme values. The framework is validated through simulation studies and real-world applications, demonstrating its robustness and improved accuracy over classical CPD methods. By addressing limitations of traditional models, this work introduces a more adaptive and statistically sound methods for detecting distributional changes in complex time series.