Hybrid Model Using Interacted-ARIMA and ANN Models for Efficient Forecasting

Abstract

When two models applied to the same dataset produce two different sets of forecasts, it is a good practice to combine the forecasts rather than using the better one and discarding the other. Alternatively, the models can also be combined to have a hybrid model to obtain better forecasts than the individual forecasts. In this paper, an efficient hybrid model with interacted ARIMA (INTARIMA) and ANN models is proposed for forecasting. Whenever interactions among the lagged variables exist, the INTARIMA model performs better than the traditional ARIMA model. This is validated through simulation studies. The proposed hybrid model combines forecasts obtained through the INTARIMA model from the dataset, and those through the ANN model from the residuals of INTARIMA, and produces better forecasts than the individual models. The quality of the forecasts is evaluated using three error metrics viz., Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE). Empirical results from the application of the proposed model on the real dataset - lynx - suggest that the proposed hybrid model gives superior forecasts than either of the individual models when applied separately. The methodology is replicable to any dataset having interactions among the lagged variables..