基于多价值链协同资源的配件价格预测模型  

Parts Price Prediction Model Based on Multi-Value Chain Collaborative Resources

作　　者：朱惠 冯玉 ZHU Hui;FENG Yu(Sichuan Key Laboratory of Manufacturing Industry Chain Collaboration and Information Support Technology,Southwest Jiaotong University,Chengdu Sichuan 611756,China;Sichuan Top Computer Vocational College,Chengdu Sichuan 611743,China)

机构地区：[1]西南交通大学制造业产业链协同与信息化支撑技术四川省重点实验室,四川成都611756 [2]四川托普计算机职业学院,四川成都611743

出　　处：《信息与电脑》2022年第10期35-37,共3页Information & Computer

摘　　要：本文面向汽车产业链协同平台上的配件代理商,以其积累的数十年销售业务数据为数据支持,提出了一种LightGBM-GRU组合预测模型。该模型利用轻量级的梯度提升机(Light Gradient Boosting Machine,LigthGBM)模型的树类模型优势,得到特征的重要性排序,选取对价格影响较大的特征集,分别带入LightGBM模型和GRU模型中,将两种模型的预测结果进行加权得到最终预测结果。选取多家代理商都销售的某多链配件进行数据预处理,并进行对比实验,结果表明该组合模型的预测结果相比于没有提取特征的单一模型的预测效果更好。This paper presents a LightGBM-GRU combination prediction model for parts agents on the automotive industry chain collaboration platform,supported by their accumulated sales business data for decades.The model takes advantage of the tree model of LightGBM model to get the importance ranking of features,selects the feature sets that have a great impact on the price,brings them into LightGBM model and Gru model respectively,and weights the prediction results of the two models to get the final prediction results.A multi chain accessory sold by many agents is selected for data preprocessing,and a comparative experiment is carried out.The results show that the prediction result of the combined model is better than that of the single model without feature extraction.

关 键 词：配件价格预测 LightGBM模型 GRU模型 组合预测模型 汽车产业链 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]