研究经济预测问题,为社会经济发展提供预测依据.由于经济时间序列是一种多维、非线性数据,采用单-的线性或非线性模型都不全面反映特点,导致预测精度不理想.为了提高经济时间序列预测精度,提出一种多变量自回归(CAR)和支持向量机(SVM)相结合的混合预测方法.混合方法首先利用CAR模型对经济时间序列的线性部分进行预测,然后采用支持向量机对非线性部分进行预测,将预测结果组合在-起,得到混合模型的预测结果.实验结果表明,混合模型的预测精度明显优于单独模型;发挥了2种模型的优势,得到一种精度高的经济预测效果.%Economic forecasting is studied in this paper. Economic time series is a kind of multi - dimensional nonlinear data, the traditional forecasting model can not fully reflect the features, and prediction accuracy is not ideal. In order to improve the economic time series forecast accuracy, a combined forecast method is put forward based on a multivariate regression (CAR) and support vector machines (SVM). Firstly, the proposed model uses CAR model to forecast the linear of economic time series, then uses support vector machine to forecast the non - linear part, and finally gets the prediction results based on the mixed model. Experimental results show that the forecast accuracy of hybrid model is much better than single model. The economic time series forecasting method is of high precision.
展开▼
