摘要
组合方法首先选取支持向量机预测算法和一阶指数平滑法对经济时间序列分别进行预测,来建立模糊自适应变权重组合预测模型。为对比模糊自适应变权重的经济时间序列组合预测模型的预测效果,选取了两种定值加权组合预测模型:平均加权模型、误差平方和最小组合预测模型。通过实验比较分析:模糊自适应变权重组合预测可以综合利用各单项预测方法的优点,比单一模型预测结果精度有了很大提高,且优于定值加权组合预测,在经济时间序列的预测方面有较高的应用价值。
This paper presents two algorithms to construct the fuzzy adaptive variable weight combination model, which are support vector machine algorithm and 1st order exponential smoothing algorithm. Firstly, the proposed model uses SVM algo- rithm and 1 st order exponential smoothing algorithm to forecast economic time series, then it is based on the predictions to construct the fuzzy adaptive variable weight combination model. For the comparation of effect this paper also is based on the predictions to construct the average weighted model, the square error and minimum combination prediction model. Result shows that the fuzzy adaptive variable weight combination model is able to integrate the advantages of every individual pre- dicting method, and compared to the monomial prediction model, the combination prediction improves the precision of the prediction greatly. Also it is superior to the constant weighted combined estimation. So it undoubtedly benefits to estimation model of economic time Serie~ nrt^clictinn_
出处
《软科学》
CSSCI
北大核心
2013年第1期141-144,共4页
Soft Science
基金
教育部人文社会科学研究青年基金项目(11YJC870002)
四川大学中央高校研究专项项目(201239)
四川大学图书情报规划项目
