Machine Learning Applications in Industry Safety: Analysis and Prediction of Industrial Accidents

This study leverages machine learning techniques to analyze industrial accident data from 12 manufacturing plants across 3 countries, with the objective of understanding the underlying patterns and predicting future accidents. By employing classification, clustering, and time-series forecasting methods, we aim to identify significant risk factors, characterize accident severity levels, and forecast accident trends. The classification model predicts accident severity with a focus on less severe incidents, while clustering analysis reveals distinct patterns related to industry sectors and specific risks. Time-series forecasting predicts a slight increase in accident counts, suggesting areas for proactive safety measures. Our findings contribute to the development of targeted safety interventions and policy formulation, emphasizing the need for industry-specific safety protocols. This research provides a foundational framework for predictive safety analytics, offering insights that could significantly enhance workplace safety management.