摘要
为了从商务响应、技术响应、价格等多个维度出发,全面评估投标人的综合实力,设计了一个基于大数据分析的电力系统供应商评估、预测方案。并选择了有代表性的算法模型,包括逻辑回归、线性回归、梯度提升树和决策树等多种模型进行了对比研究。通过对某省级电力公司四年的历史数据进行分析,算例显示:梯度上升模型在多种特征下均显示了较高的接近90%的正确率,而逻辑回归模型在f4特征下得到了最高正确率91.86%,从而在一定程度上证明了利用大数据分析手段对电力供应商进行投标评估、预测的可能性和可行性。
In order to comprehensively evaluate the overall strength of bidders,this paper designed a power system supplier evaluation and forecasting program with dimensions such as business response,technical response and price based on big data analysis.Some representative algorithm models,including logistic regression,linear regression,gradient elevating trees,and decision trees,were selected for the comparative study.A four-year historical dataset of a provincial power company is taken as the researching example.The result shows that the gradient-rising model gives a high accuracy of nearly 90%under various characteristics,while the logistic regression model obtains the highest accuracy 91.86%under a certain feature,which proves,in a certain degree,the possibility and feasibility of applying big data analysis to evaluate and predict the bidding of electric power suppliers.
作者
魏俊奎
周顺凯
金义
WEI Junkui;ZHOU Shunkai;JIN Yi(National Grid Anhui Electric Power Co.,Hefei 230022;Department of Computer Science and Technology,East China Normal University,Shanghai 200062)
出处
《微型电脑应用》
2019年第10期75-78,共4页
Microcomputer Applications
关键词
投标
机器学习
大数据
Bid
Machine learning
Big data
