Early warning method for the commodity prices based on artificial neural networks: SMEs case

Applications based on Artificial Neural Networks (ANN) have been developed thanks to the advance of the technological progress which has permitted the development of sales forecasting on consumer products, improving the accuracy of traditional forecasting systems. The present study compares the performance of traditional models against other more developed systems such as ANN, and Support Vector Machines or Support Vector Regression (SVM-SVR) machines. It demonstrates the importance of considering external factors such as macroeconomic and microeconomic indicators, like the prices of related products, which affect the level of sales in an organization. The data was collected from a group of supermarkets belonging to the SMEs sector in Colombia. At first, a pre-processing was carried out to clean, adapt and standardize data bases. Then, since there was no labeled information about the pairs of substitute or complementary products, it was necessary to implement a cross-elasticity analysis. In addition, a harmonic average (f1-score) was considered at several points to establish priorities in some products and obtained results. The model proposed in this study shows its potential application in the product sales forecasting with high rotation in SMEs supermarkets since their results are more accurate than those obtained using traditional procedures.

• Autores: Silva, Jesus; MOJICA HERAZO, JULIO CESAR; Rojas Millán, Rafael Humberto; Pineda Lezama, Omar Bonerge; Morgado Gamero, W.B.; Varela Izquierdo, Noel
• Tipo de recurso: Article of journal
• Fecha de publicación: 2019
• Institución: Corporación Universidad de la Costa
• Repositorio: REDICUC - Repositorio CUC
• Idioma: eng
• OAI Identifier: oai:repositorio.cuc.edu.co:11323/4840
• Acceso en línea: http://hdl.handle.net/11323/4840; https://repositorio.cuc.edu.co/
• Palabra clave: predictive model; Multilayer Perceptron; Multiple input multiple output; Forecast; Support vector machines; Cyclic variation
• Rights: openAccess
• License: Attribution-NonCommercial-NoDerivatives 4.0 International