Fuzzy forecasting based on automatic clustering and axiomatic fuzzy set classification

In spite of the impressive diversity of models of fuzzy forecasting, there is still a burning need to arrive at models that are both accurate and highly interpretable. This study proposes a new fuzzy forecasting model designed with the use of the two key techniques, namely clustering and axiomatic fuzzy set (AFS) classification. First, clustering algorithm is utilized to generate clustering-based intervals. Second, the fuzzy trend labeled training data set is constructed based on fuzzy logic relationships and fuzzy trends of historical samples. Then, the AFS classification is exploited to yield the semantic interpretation of each fuzzy trend. The main novelty is that the proposed model not only predicts the value but can also capture the trend prevailing in the time series, and obtain its semantic interpretation. The Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX), inventory demand, and Spanish electricity prices are used in a series of experiments. The results show that the proposed model has both good interpretability and accuracy.