Determination of Fuzzy Relations for Economic Fuzzy Time Series Models by Neural Networks

keywords: Fuzzy time series models, fuzzy bank rules, product-space clustering, fuzzy controllers, neural networks
Based on the works citekosko2, song, zadeh a fuzzy time series model is proposed and applied to predict chaotic financial process. The general methodological framework of classical and fuzzy modelling of economic time series is considered. A complete fuzzy time series modelling approach is proposed which includes: determining and developing of fuzzy time series models, developing and calculating of fuzzy relations among the observations, calculating and interpreting the outputs. To generate fuzzy rules from data, the neural network with SCL-based product-space clustering is used.
reference: Vol. 22, 2003, No. 5, pp. 457–471