Produção industrial capixaba : uma análise comparativa dos principais métodos estatísticos de combinação de previsão
Ano de defesa: | 2018 |
---|---|
Autor(a) principal: | |
Orientador(a): | |
Banca de defesa: | |
Tipo de documento: | Dissertação |
Tipo de acesso: | Acesso aberto |
Idioma: | por |
Instituição de defesa: |
Universidade Federal do Espírito Santo
BR Mestrado em Economia Centro de Ciências Jurídicas e Econômicas UFES Programa de Pós-Graduação em Economia |
Programa de Pós-Graduação: |
Não Informado pela instituição
|
Departamento: |
Não Informado pela instituição
|
País: |
Não Informado pela instituição
|
Palavras-chave em Português: | |
Link de acesso: | http://repositorio.ufes.br/handle/10/8783 |
Resumo: | The objective of this work is to specify forecast models and forecast combinations applied to the time series of indices in the Extraction Industry, Transformation Industry and General Industry for the State of Espírito Santo. A priori, it focused on evaluating the performance of the main forecasting methods available in the literature, such as Holt-Winters, Box-Jenkins, Models of Artificial Neural Networks, and Econometric Models, incorporating other econometric variables in the latter method. such as inflation, interest rate, unemployment rate, industrial entrepreneur confidence index, utilization of installed capacity, among others. Secondly, it is considered to select the best model estimated for each methodology, and then to apply predictive combination methods, in order to evaluate if there is a difference between the accuracy of individual forecasts and their combinations. As technique combination forecast, were considered the methods of arithmetic mean, simplified minimum variance, and ordinary least squares regression. The performance evaluation of predictions and combinations of forecasts is obtained by accuracy measurements know by MAE, MSE, RMSE, MAPE, SMAPE, and U - Theil's. As the main result obtained, we highlight the predictions obtained from the forecast combination method ordinary least squares, which unanimously performed better than the other predictions for the three industrial production series considered in this study. |