Citation

Özkan EK, Ulaş HB. Safety Sci. 2024; 174: e106468.

Copyright

(Copyright © 2024, Elsevier Publishing)

DOI

10.1016/j.ssci.2024.106468

PMID

unavailable

Abstract

Occupational accidents are one of the main problems in production system especially in metal sector. The aim of this study is to develop a predictive framework using machine learning (ML) to identify the causes of fatalities and amputations in the metal sector based on occupational accident data collected by the Turkish Ministry of Labor and Social Security (MLSS) from 2013 to 2018. Researchers have used a variety of strategies to investigate factors and create effective prediction frameworks for lowering occupational accidents. We used random forest (RF), k-nearest neighbour (KNN), gradient boosting method (GBM) and recursive partitioning and regression trees (RPART) to predict accident causes and consequence. Accuracy, precision, recall and f-score is used to measure the performance of ML frameworks. For model validation 10-fold cross validation method was used which increased the accuracy of the frameworks considerably. We extracted important factors which affected the causes of accident at metal sector using feature importance. Analysis proved RF as the best performing framework with highest classification results with 0.9172 accuracy, 0.9618 precision, 0.9518 recall and 0.9568 f-score using all features as compared to other techniques classification of occupational accident severity. To implement preventive controls and interventions in a more targeted way, it is recommended to use the predictive RF algorithm in the analysis of occupational accidents. With these studies, preventive measures can be taken by predicting occupational accidents that may occur in the future.

Language: en

Keywords

Gradient Boosting Method; K-nearest Neighbour; Occupational Accident; Random forest; Recursive Partitioning and Regression Trees