Análise das taxas de mortalidade por acidente de trabalho na região sul do Brasil

• Autor(a) principal: Melchior, Cristiane
• Data de Publicação: 2019
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Manancial - Repositório Digital da UFSM
• Texto Completo: http://repositorio.ufsm.br/handle/1/20636
• topic: Taxa de mortalidade; Acidente de trabalho; ARIMA; Previsão; Séries temporais; βARMA; Mortality rate; Work accident; Beta distribution; Prediction; Time series; CNPQ::ENGENHARIAS::ENGENHARIA DE PRODUCAO

Fatal occupational accident rates are explored in order to measure workers' exposure to the risks inherent to the economic activity, making it possible to monitor the oscillations and the historical trend of work-related fatal accidents, as well as the impacts on organizations and society. Considering the relevance of the topic, this study intends to use the ARIMA and ��ARMA methodologies to model the series of data of fatal work accident rates, to compare the adjusted prediction models and to identify the best predictive model by state of the southern region of Brazil (Paraná (PR), Santa Catarina (SC) e Rio Grande do Sul (RS)). The labor-related fatality rates were obtained by the Department of Information Technology of SUS (DATASUS), by means of deaths recorded in the Mortality Information System (SIM), stratifying them by state of the southern region of Brazil. With this information a descriptive analysis was carried out, showing the vulnerability characteristics of the victims, comparing them among the three states. Also, three time series were modeled using the Box-Jenkins or ARIMA and ��ARMA methodologies, for the monthly period from 2000 to 2016. Finally, it was evident that there was an increase in female participation in the labor market, but the number of deaths from occupational accidents did not increase in the same proportion. The state of Paraná stood out having the highest mortality rate due to work accidents. A better performance in the adjustment of the ��ARMA model was observed in the 6-month time horizon using the MAE, MAPE and RMSE accuracy measurements, the residue analysis and the AIC and BIC penalty criteria to evaluate the quality of the models and select the most accurate forecast model