ARIMA Neural Network Research Articles

Trading and forecasting performance of different hybrid ARIMA - neural network models for stock returns

This study examines the performance of different hybrid methodologies that combine ARIMA and artificial neural network (ANN) to forecast stock market returns. Two new hybrid ARIMA-ANN models are developed and compared with Zhang’s (2003) model on real data sets and the model performance is evaluated using trading performance measures. The study shows that hybrid models outperform independent models and the hybrid ARIMABP model achieves greater accuracy and provides evidence of superiority of the other hybrid models.

Evaluating ARIMA-Neural Network Hybrid Model Performance in Forecasting Stationary Timeseries

Demand prediction is one of most sophisticated steps in planning and investments. Although many studies are conducted to find the appropriate forecasting models, dynamic nature of forecasted parameters and their effecting factors are apparent evidences for continuous researches. ARIMA, Artificial Neural Network (ANN), and ARIMA-ANN hybrid model are well-known forecasting models. Many researchers concluded that the Hybrid model is the predominant forecasting model in comparison with ARIMA and ANN individual models. Most of these researches are based on non-stationary or seasonal timeseries, whereas in this article, hybrid models forecast ability by stationary time series is studied. Some following demand time steps from a paint manufacturing company are forecasted by all previously mentioned models and ARIMA-ANN hybrid model fails to present the best forecasts.

Time-series analysis with neural networks and ARIMA-neural network hybrids

Time-series analysis is important to a wide range of disciplines transcending both the physical and social sciences. Statistical models have sound theoretical bases and have been successfully used in a number of problem domains. More recently, machine-learning models such as neural networks have been suggested as offering potential for time-series analysis. Results of neural network empirical testing have thus far been mixed. This paper proposes melding useful parameters from the statistical ARIMA model with neural networks of two types: multilevel perceptrons (MLPs) and radial basis functions (RBFs). Tests are run on a range of time-series problems that exhibit many common patterns encountered by analysts. The results suggest that hybrids of the type proposed may yield better outcomes than either model by itself.