Abstract
Workers’ health and safety are a major concern in society, since work accidents have a major impact on productivity and economy. In Brazil, the accidents are officially reported through Work Accident Communication and they are available to the public. Thus, this study analyzed a balanced dataset containing 1,206 records of deaths caused by work accidents related to the transport sector. Its aim was analyzing how the deaths in the transport sector are related with the other work accident factors. To achieve this goal, twelve performance data mining techniques are compared, through five performance metrics, regarding the predictive capacity of the occurrence of deaths caused by work accidents. In this context, the XGBoost and Naïve Bayes algorithms showed the best predictive capacity. The explanatory analysis indicates that work accidents followed by death in road transport are predictable due to the severity of the injuries and vital parts of the body are affected.
Publisher
Universidad Nacional de Colombia
Reference38 articles.
1. Associação Brasileira de Normas Técnicas. NBR 14280: Cadastro de acidente do trabalho - Procedimento e classificação. Rio de Janeiro: ABNT, 2001.
2. Batista, A.G., Santana, V.S. e Ferrite, S., Registro de dados sobre acidentes de trabalho fatais em sistemas de informação no Brasil. Ciência & Saúde Coletiva, 24, pp. 693-704, 2019. DOI: https://doi.org/10.1590/1413-81232018243.35132016
3. Berson, A. and Thearling, K., Building data mining applications for CRM. McGraw-Hill, Inc., USA, 1999.
4. Bevilacqua, M., Ciarapica, F.E. and Giacchetta, G., Industrial and occupational ergonomics in the petrochemical process industry: a regression trees approach. Accident Analysis & Prevention, 40(4), pp. 1468-1479, 2008. DOI: https://doi.org/10.1016/j.aap.2008.03.012
5. Cabena, P., Hadjinian, P., Stadler, R., Verhees, J. and Zanasi, A., Discovering data mining: from concept to implementation. Prentice-Hall, Inc., USA, 1989.