1Universidad Federal de Pernambuco, Brasil, Maestro en Estadística. E-mail: csantana@cable.net.co
Palabras Claves: Perceptron multicapas, modelos SARIMA, suavizamiento exponencial, combinación de pronósticos, componentes no observables.
Evaluating the usefulness of neural network methods in predicting the Colombian Inflation is the main goal of this paper. The results show that neural networks forecasts can be considerably more accurate than forecasts obtained using exponential smoothing and SARIMA methods. Experimental results also show that combinations of individual neural networks forecasts improves the forecasting accuracy.
Key words: Multilayer perceptron, SARIMA models, Exponencial smooth- ing, Combination of forecasts, Unobservable components.
Texto completo disponible en PDF
Referencias
1. Aristizábal, M. & Misas, M. (2006), Evaluación asimétrica de una red neuronal artificial: una aplicación al caso de la inflación en Colombia, Technical report, Working Paper 377. Subgerencia de estudios económicos–Banco Central de Colombia. [ Links ]
2. Barnard, G. A. (1963), "New Methods of Quality Control", Journal of the Royal Statistical Society A 126, 255–259. [ Links ]
3. Box, G. E. P. & Jenkins, G. M. (1976), Time Series Analysis: Forecasting and Control, San Francisco: Holden-Day. [ Links ]
4. Box, G. E. P., Jenkins, G. M. & Reinsel, G. (1994), Time Series Analysis: Fore- casting and Control, 3a. edn, Englewood Cliffts: Prentice Hall. [ Links ]
5. Cohen, M., Franco, H., Morgan, N., Rumelhart, D. & Abrash, V. (1993), Ad vances in Neural Information Processing Systems, Morgan Kaufmann, chapter Context-dependent multiple distribution phonetic modeling with MLPs, pp. 649–657. [ Links ]
6. Cybenko, M. (1989), "Aproximation by Superposition of a Sigmoidal Function", Mathematics of Control, Signals and Systems 2, 303–314. [ Links ]
7. Gately, E. (1996), Neural Networks for Financial Forecasting, New York: John Wiley and Sons. [ Links ]
8. Gómez, V. & Maravall, A. (1996), Programs Tramo (Time series Regression with Arima noise, Missing observations, and Outliers) and Seats (Signal Extraction in Arima Time Series). Instructions for the User, Technical report, Working Paper 9628. Servicio de estudios-Banco de España. [ Links ]
9. Haykin, S. (1994), Neural Networks, New York: McMillan College Publishing Company. [ Links ]
10. Hendry, D. F. & Clements, M. P. (2004), "Pooling of Forecast", Econometrics Journal 7, 1–31. [ Links ]
11. Hornik, K., Stinchcombe, M. & White, H. (1989), "Multilayer Feedforward Networks and Universal Approximations", Neural Networks 2, 359–366. [ Links ]
12. Isasi, P. & Galván, I. (2004), Redes neuronales artificiales – un enfoque práctico, Madrid: Pearson-Prentice Hall. [ Links ]
13. Jalil, M. & Misas, M. (2006), Evaluación de pronósticos del tipo de cambio utilizando redes neuronales y funciones de pérdida asimétricas, Technical report, Working Paper 376. Subgerencia de estudios económicos–Banco Central de Colombia. [ Links ]
14. Kaastra, I. & Boyd, M. (1996), "Design a Neural Network for Forecasting Financial and Economic Time Series", Neurocomputing 10, 215–236. [ Links ]
15. Maravall, A. & Kaiser, R. (2000), Notes on Time Series Analysis, ARIMA models and Signal Extraction. *http:www.bde.es/servicio/software/trabajos.htm [ Links ]
16. McLeod, A. I. & Li, W. K. (1983), "Diagnostic Checking ARMA Time Series Models Using Squared-Residuals Autocorrelation", Journal of the Time Series Analysis 4, 269–273. [ Links ]
17. Misas, M., López, E., Arango, C. & Hernández, J. (2003), La demanda de efectivo en Colombia: una caja negra a la luz de las redes neuronales, Technical report, Working Paper 268. Subgerencia de estudios económicos–Banco Central de Colombia. [ Links ]
18. Misas, M., López, E. & Borrero, P. (2002), "La inflación en Colombia: una aproximaci ón desde redes neuronales", Ensayos sobre Política Económica 41- 42, 143–214. [ Links ]
19. Montgomery, D. C. & Johnson, L. A. (1976), Forecasting and Time Series Analysis, New York: McGraw-Hill. [ Links ]
20. Morettin, P. A. & Toloi, C. M. (2004), Análise de Séries Temporais, ABE - Projeto Fisher. São Paulo: Edgard Blücher. [ Links ]
21. Narendra, K. & Parthasaranty, K. (1990), "Identification and Control of Dynamical Systems Using Neural Networks", IEEE Transactions on Neural Networks 1, 4–27. [ Links ]
22. Ospina, R. M. & Zamprogno, B. (2003), "Compara¸cão de Algumas Técnicas de Previsão em Análise de Séries Temporais", Revista Colombiana de Estadística 26(2), 129–157. [ Links ]
23. R Development Core Team (2005), R: A language and environment for statistical computing, R Foundation for Statistical Computing, Vienna, Austria. *http://www.R-project.org [ Links ]
24. Rumelhart, D., Hilton, G. & Williams, R. (1986a), "Learning Representations by Backpropagating Errors", Nature 323, 533–536. [ Links ]
25. Rumelhart, D., Hilton, G. & Williams, R. (1986b), The MIT Press,chapter Learning representations by backpropagating errors, Parallel Distributed Processing, Cambridge. [ Links ]
26. Souza, R. C. & Zandonade, E. (1993), Forecasting Via Neural Networks: Comparative Study, Technical report, Catholic University of Rio de Janeiro, Department of Electrical Engineering. [ Links ]
27. Varfis, A. & Versino, C. (1990), Univariate Economic Time Series Forecasting, Cambridge University Press. [ Links ]
28. Wieggend, A., Huberman, B. & Rumelhart, D. (1990), Predicting the Future: a Connectionist Approach, Technical report, PARC. [ Links ]
29. Zhang, G., Patuwo, B. & Hu, Y. (1998), "Forecasting with Artificial Neural Networks: The State of Art", International Journal of Forecasting 14, 35–62. [ Links ] ]]>