K-nearest neighbor LS-SVM method for multi-step prediction of chaotic time series

To reduce the complexity of training an least squares support vector machine (LSSVM), a nearest neighbors method was proposed to perform Multi-step time series prediction. By selecting dataset with smallest Euclidean distance and similar changing trend for each testing instance, a reduced training dataset was defined. Experiments on chaotic datasets were conducted to compare the prediction performance with traditional interactive single-step methods. The results demonstrate that the proposed method outperforms the single-step methods. The ability of Multi-step prediction is promising even when the noises were added.