摘要
针对线性回归和多层感知器(multi-layer perceptron,MLP)神经网络的理论与运用方法进行研究,利用SPSS软件搭建线性回归模型和MLP神经网络预测模型,通过招标采购实际预测算例,引入平均相对误差和标准误差两种指标对模型的预测精度进行比对。结果表明,对于产品价格和原材料价格大致呈线性关系的变电设备,两种模型预测精度无明显差别,在招标采购价格预测实际应用中均具有一定的参考价值。
The mechanism and function pattern of linear regression and multi-layer perceptron(MLP)neural network were discussed.These two models were built by SPSS software on basis of actual statistics to forecast the bid price.The mean relative error and standard error were introduced to compare the prediction accuracy of the two models through an example of bidding and procurement.The results show that the product prices and raw materials prices of the transformation equipments roughly appear liner relation,the forcasting precision of the two models has no obvious difference,which both have certain reference value in the practical application of bidding procurement price prediction.
作者
庞铖铖
戎袁杰
刘昕
宋梦昕
王光旸
PANG Chengcheng;RONG Yuanjie;LIU Xin;SONG Mengxin;WANG Guangyang(State Grid Materials Co.,Ltd.,Beijing 100120,China)
出处
《宁夏电力》
2021年第1期12-17,共6页
Ningxia Electric Power
关键词
招标采购预测
价格预测
线性回归
多层感知器
神经网络
bidding procurement forcasting
price forecasting
linear regression
MLP
neural network
