Paper aims: In this study, effective strategies to combine and select forecasting methods are proposed. In the selection strategy, the best performing forecasting method from a pool of methods is selected based on its accuracy, whereas the combination strategies are based on the mean methods' outputs and on the methods' accuracy. Originality: Despite the large amount of work in this area, the actual literature lacks of selection and combination strategies of forecasting methods for dealing with intermittent time series. Research method: The included forecasting methods are state-of-the-art approaches applied to industrial and academics forecasting problems. Experiments were performed to evaluate the performance of the proposed strategies using a spare part data set of an industry of elevators and a data set from the M3-Competition. Main findings: The results show that, in most cases, the accuracy of the demand forecasts can be improved when using the proposed selection and combination strategies. Implications for theory and practice: The proposed methodology can be applied to forecasting problems, covering a variety of characteristics (e.g., intermittency, trend). The results reveal that combination strategies have potential application, perform better than state-of-the-art models, and have comparable accuracy in intermittent series. Thus, they can be employed to improve production planning activities.
Previsões de séries temporais auxiliam a tomada de decisão em diversas áreas como marketing, economia e indústria, sendo que a principal finalidade é estimar o comportamento futuro de uma sequência de observações. Nesse sentido, conjuntos de modelos (ensembles) híbridos, que combinam modelos de aprendizado de máquina e estatísticos, têm se mostrado eficientes para prever séries temporais. Entretanto, a correta seleção dos modelos e da combinação em um ensemble é importante para assegurar o desempenho do sistema. Assim, este trabalho propõe e compara diferentes abordagens de ensembles híbridos para melhorar a previsão de séries temporais. São propostos um conjunto de modelos de previsão e diferentes estratégias de combinação para lidar com séries temporais de diferentes padrões. A primeira abordagem de ensembles combina um conjunto de modelos com acurácia utilizando quatro estratégias de combinação. Já a segunda abordagem ensemble seleciona automaticamente modelos e a combinação utilizando meta-heurísticas. As abordagens são comparadas utilizando um novo conjunto de dados de uma empresa de distribuição de cosméticos e um conjunto de dados público. Os resultados demonstram que os ensembles propostos são eficientes para diminuir o erro de previsão.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.