Unsupervised Machine Learning for Pattern Identification in Occupational Accidents

Abstract

Creating safe work environment is significant in saving workers’ lives, improving corporates’ social responsibility and sustainable development. Pattern identification in occupational accidents is vital in elaborating efficient safety counter-measures aiming at improving prevention and mitigating outcomes of future incidents. The objective of this study is to identify patterns related to the occurrence of occupational accidents in non-farm agricultural work environments based on workers’ compensation claims data, using latent class clustering method as an un-supervised machine learning modeling approach. The result showed injury profiles and incident dynamics have low, average, and high levels of risks based on the main causes and outcomes of the injuries and the affected body part(s).

Keywords: Occupational Accidents, Unsupervised Machine Learning, Data Analytics, Safety Analytics

DOI: 10.54941/ahfe1001089