A Novel Hybrid Model for Time Series Forecasting Using Artificial Neural Network and Autoregressive Integrated Moving Average Models

Enhancing forecast accuracy while using time series is a potential area of research. Evidences exist in the literature to show that hybrid models can significantly improve the forecasting performance, as they combine the exclusive strengths of different models. This paper presents a novel hybrid model by combining forecasts from Autoregressive Integrated Moving Average (ARIMA) and artificial neural network (ANN) models with suitable weights, thereby improving the forecast accuracy. The methodology employs appropriate error metrics to construct the weights. The paper further demonstrates the efficiency of the proposed methodology through an empirical study, based on two real-world time series data sets. Thus, the new methodology can be used for enhancing the forecast accuracy in a number of fields of research.