传统的集对预测方法大多只能作短期预测,而且多数属于线性模型,没有考虑蕴含在同异反联系度中的非线性、时变性和强耦合关系.本文通过Logratio变换与反变换,将支持向量回归方法用于集对预测,在一定程度上可以克服线性建模技术的不足.该模型被用于人力资源绩效预测,取得了较好的效果.
The traditional method of set-pair analysis can only make short-term predictions and mostly belongs to the linear model; it doesn't take into consideration the nonlinearity, coupling and dynamicity embedded in the homology, difference and opposition properties of set-pairs. This paper applies SVR to the prediction of set-pairs through transformation and countertransformation of Logration and overcomes the shortcomings of linear models. It is used to predict the performance of human resources and shows good results.