This article aims to study the intermittent demand forecasting for a specific type of spare part of a Brazilian refrigeration industry that commercialize its products in the Latin American market. Demand characterization is performed in terms of their intermittency and variability. Results are obtained with classical intermittent forecasting methods outside the sample: Croston, Syntetos-Boylan Approximation (SBA), Shale-Boylan-Johnston Correction (SBJ), Multiple Aggregation Prediction Algorithm (MAPA) and with Artificial Neural Networks (ANN) based models. Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) are used for comparison and selection of forecast model. The comparative analysis results shows that the predictions based on a simple three-layer ANN model trained with the Resilient Backpropagation algorithm present better performance. The calculations were performed using R software with RStudio, "forecast", "tsintermittent" and "neuralnet" libraries.Keywords: demand forecasting; intermittent demand; artificial neural networks.Previsión de demanda intermitente con métodos de series de tiempo y redes neuronales artificiales: Estudio de caso ResumenEste artículo tiene como objetivo estudiar la previsión de la demanda intermitente de un tipo específico de pieza de reposición en una industria brasilera de sistemas de refrigeración que comercializa sus productos en el mercado latinoamericano. La demanda es caracterizada en términos de intermitencia y variabilidad. Los resultados son calculados usando métodos clásicos de previsión intermitente fuera de la muestra: Croston, Aproximación Syntetos-Boylan (SBA), Corrección Shale-Boylan-Johnston (SBJ), Algoritmo de Previsión de Agregación Múltiple (MAPA) y modelos basados en Redes Neuronales Artificiales (RNA). El Error Cuadrático Medio (RMSE) y Desvío Medio Absoluto (MAE) son utilizados para efectos de comparación y selección del modelo de previsión. El análisis comparativo de los resultados muestra que las previsiones basadas en modelos RNA simple de tres capas y entrenadas con el algoritmo Resilient Backpropagation presentan mejor desempeño. Los cálculos fueron realizados con el software R, RStudio, bibliotecas "forecast", "tsintermittent" y "neuralnet".Palabras clave: previsión de la demanda; demanda intermitente; redes neuronales artificiales.
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.