一种模糊组合预测方法

Method of Fuzzy Combination Forecasting

针对预测问题的不确定性,提出一种将组合预测结果扩展到置信集合的模糊组合预测方法.该方法利用概率中置信区间原理并结合模糊数概念,对观测值和各单项预测值进行模糊描述;以组合预测值对观测值的拟合度最高为目标,建立模糊规划模型,求解权重系数.该组合预测方法与单项预测方法相比,预测精度有所提高,具有一般组合预测方法的优点,并以模糊数的形式将预测结果确定在一置信集合内.最后以实例证明了该方法的有效性.

Aimed against the uncertainty in forecasting problems, a method of fuzzy combination forecasting is proposed whereby the outcome of combination forecasting is expanded to a confidence interval. By utilizing the theory of confidence interval and the concept of fuzzy number, the outcome of observation and forecasting are transformed into fuzzy numbers. Aimed at maximizing the imitation degree, a model of fuzzy programming is set up to calculate the combination-weighted coefficient. The method not only has the advantages that other ways of combination forecasting share, but it also expands the outcome of forecasting to a confidence interval, which is proved to be effective by an example.