Technical Analysis of Occupational Fatal Accidents in Malaysia Using Machine Learning Techniques

抄録

With the rapid economic growth of Malaysia, workplace accidents have increased drastically, according to the Department of Occupational Safety and Health (DOSH). This study aimed to determine the patterns in Malaysian workplace fatal accidents. A total of 505 fatal accident cases across 15 industries were analyzed in this study using both qualitative and quantitative methods. These fatality cases were identified and recorded by the DOSH from 2010 to 2020. The data were arranged and coded in Python and analyzed in terms of frequency analysis, Spearman's rank order correlation, eta squared, chi-square, and Cramer's V methods. Furthermore, neuro-linguistic programming was performed for word cloud and sentiment analyses. Finally, a light gradient-boosting machine learning model was used to further understand the causes of fatalities in Malaysia. The results showed that fatal falls from heights were the highest contributor to fatal accidents (32%, n = 161). Workers under contract were more vulnerable to fatal accidents in the construction industry (n = 324, 64%) than other workers. General workers were the most susceptible category to fatal accidents (60%, n = 302). The results from this study provide valuable insights into workplace fatal accident patterns and strategies for their prevention across industries.