Fuzzy Time Series Prediction with Data Preprocessing and Error Compensation Based on Correlation Analysis

In general, it is difficult to predict non-stationary or chaotic time series since there exists drift and/or non-linearity as well as uncertainty in them. To overcome this situation, we propose an effective prediction method which adopts data preprocessing and multiple model TS fuzzy predictors combined with model selection mechanism. The proposed method uses the differences of time series as predictor input instead of the original ones because the difference data can stabilize the statistical characteristics of those time series and reveals better their implicit properties. In data preprocessing procedure, the candidates of optimal difference interval are determined based on the correlation analysis, and corresponding difference data sets are generated for them. And then, TS fuzzy predictors are constructed for multiple model bank, where k-means clustering algorithm is used for fuzzy partition of input space, and the least squares method is applied to parameter identification of fuzzy rules. Among the predictors in the model bank, the best one which minimizes the performance index is selected and it works on hereafter for prediction. Finally, the error compensation procedure based on correlation analysis is added to improve the prediction accuracy. Computer simulation on some typical examples is performed to verify the effectiveness of the proposed method.