论文部分内容阅读
传统的金融时间序列预测方法以精确的输入数据为研究对象,在建立回归模型的基础上做单步或多步预测,预测结果是一个或多个具体的值.由于金融市场的复杂性,传统的预测方法可靠度较低.提出将金融时间序列模糊信息粒化成一个模糊粒子序列,运用支持向量机对模糊粒子的上下界进行回归,然后应用回归所得到的模型分别对上下界进行单步预测,从而将预测的结果限定在一个范围之内.这是一种全新的思路.以上证指数周收盘指数为实验数据,实验结果表明了这种方法的有效性.
The traditional method of forecasting financial time series takes the accurate input data as the research object and makes one-step or multi-step forecast based on the regression model, the forecast result is one or more specific values.As a result of the complexity of financial markets, the traditional The proposed method can be used to regress the upper and lower bounds of the fuzzy particle by using the support vector machine to model the fuzzy information of the financial time series into a fuzzy particle sequence and then to predict the upper and lower bound separately by applying the regression model , So as to limit the forecast results within a range.This is a new way of thinking.With the closing index of the Shanghai Composite Index Week as experimental data, the experimental results show the effectiveness of this method.