摘要
为了准确、快速地对铁路物流需求量进行预测,针对现有铁路物流需求量预测模型存在的问题,采用梯度提升算法对分类与回归树算法进行集成,提出一种GB-CART集成算法。以1990-2014年的铁路物流需求量为研究对象,选取预测年份前3年的铁路物流需求量作为模型输入,预测年份铁路物流需求量作为模型输出,采用GBCART集成算法进行仿真实验,并与单一CART、SVR、RBF和LR模型进行比较。结果表明：GB-CART模型的预测效果与单一CART模型相比得到了大幅度提升,且预测精度高于SVR、RBF和LR,验证了所提出模型的有效性及准确性。
In order to forecast the railway logistics demand accurately and quickly, a GB-CART ensemble algorithm is pro- posed,which is focus on the problems of the existing railway logistics demand forecasting model. The GB-CART ensemble algorithm is built through using the grandient algorithm to integrate the classification and regression tree algorithm. The rail- way logistics demand during 1990 - 2014 is selected as the research object. The railway logistics demand of three years be- fore the forecast year are selected as the input of the model, and the railway logistics demand of the forecast year is selected as the output of the model, the GB-CART ensemble algorithm is used to carry out simulation experiment. At the same time, the resuhs of GB-CART ensemble algorithm is compared with the results of single CART, SVR, RBF and LR model. The re- sults show that the forecast effect of GB-CART model has been greatly improved compared with the single CART model, and the prediction accuracy is higher than that of SVR, RBF and LR, which verifies the validity and accuracy of the proposed model.
出处
《世界科技研究与发展》
CSCD
2016年第6期1311-1314,共4页
World Sci-Tech R&D
基金
国家自然科学基金(51374121)资助
关键词
铁路物流需求量
集成学习
分类与回归树
梯度提升
railway logistics demand
ensemble learning
classification and regression tree
grandient boosting
