Study on Adaptive Hybrid Prediction Algorithm and its Implementation Based on Neural Network

Abstract

In this paper, nonlinear time series forecasting system combining algorithm proposed prediction model. For the model of the existing combination forecasting method selection and mixed results so that it can be improved terms for a variety of different sequences with adaptive prediction. The results show that for different test data set, the method can effectively use all kinds of prediction Models pool without specific filter to adjust the mixing weight ratio of each of the prediction results so that the adaptive prediction, ensure higher prediction accuracy achieved.