A Hybrid Forecasting Model Based on Adaptive Fuzzy Time Series and Particle Swarm Optimization

Limited historical data and large fluctuations are two important issues for forecasting time series. In this paper, a hybrid forecasting model based on adaptive fuzzy time series and particle swarm optimization is proposed to address these issues. In the training phase, the heuristic rules automatically adapt the forecasted values based on trend values and the particle swarm optimization is applied to adjust the interval lengths in the universe of discourse for accuracy forecasting. The root mean square error, the absolute percent error and the mean absolute percent error are used to evaluate the forecasting performance. The data of tourism from Taiwan to the United States are used in the empirical study. The experimental results show that the proposed forecasting model outperforms other listed models in both the training and testing phases.