Advances and Applications in Statistics
Volume 37, Issue 1, Pages 37 - 71
(November 2013)
|
|
HYBRID MODEL OF MARIMA AND ANNBP FOR FORECASTING SALES TAX IN EGYPT
Medhat M. A. Abdelaal, Moustafa Galal Moustafa and Dalia Mohamed Samy Aglan
|
Abstract: This article proposes a hybrid forecasting model, which combines the multivariate time series MARIMA and the artificial neural network (ANN) using back propagation (BP) technique, and hybrid model that combines multivariate MARIMA with ANN known as MARIMABP is developed to improve the forecast accuracy. First, MARIMA model was used to model the linear component and then a neural network model was developed to model the residuals from the MARIMA model using monthly time series of sales tax of seven products in Egypt from Jan. 2003 to June 2013. This monthly time series was divided into two parts: the first part from Jan. 2003 to Dec. 2010 and the second part from Jan. 2011 to June 2013. The first part used to build the models and the second part used as a validation part to check the validity of the suggested models resulted from the first part. This model was used to predict 6 months of sales tax data for seven products in Egypt from July 2013 to Dec. 2013. The forecasting performance was compared among three models: multivariate MARIMA, the ANN (BP) and MARIMABP models. Among these methods, the mean square error (MSE), the mean absolute error (MAE), root mean square error (RMSE), and the mean absolute percentage error (MAPE) of the MARIMABP model were the lowest, the mean percentage error (MPE) was very close to zero and the coefficient of determination was the highest. The results indicated that hybrid model can be an effective way to improve forecasting accuracy comparing with the single MARIMA model and the single ANN model. |
Keywords and phrases: ARIMA, multivariate MARIMA, artificial neural networks, hybrid model, backpropagation, MARIMABP, connection between neurons, transfer functions, learning parameters. |
|
Number of Downloads: 362 | Number of Views: 1419 |
|