A Forecasting Model based on High-Order Fuzzy Time Series and Simulated Annealing Algorithm, Case Study: Tehran's Bourse Price Index (TEPIX)

During the recent years extensive researchs have been done on fuzzy time series. Since length of intervals affect the forecasting results in these models, doing research in this area became an interesting topic for time series researchers, there are some studies on this issue but their results are not good enough.In this study, we propose a novel simulated annealing heuristic algorithm is used to promote the accuracy of forecasting. The algorithm enjoys two new neighborhood search operators called “subtitle” and “adjust". A Taguchi method as an optimization technique is also employed to comprehensively tune different parameters and operators of proposed model. The experimental results show that proposed models (SAFTS(is more accurate than existing models on forecasting Alabama university enrollments. At the final step, Tehran’s bourse price index (TEPIX) was used as a case study for forecasting and results indicate a good forecasting performance on this problem.