摘要
提出通过信息熵构造判定树的数据挖掘算法对历史竞标样本进行分类的新思路。介绍了算法中如何处理高分枝属性、数值属性和缺失数据及剪枝等关键环节。一个考虑市场需求水平、机组报价水平、机组容量等多因素的算例说明了该算法的实现过程并得到机组的负荷率与这些因素一些潜在的规则性知识 ,从而得到不同特征机组在市场中的竞标能力。
A new data-mining framework for competitive bidding assessment in deregulated power market is proposed in this paper. Some key aspects about algorithm are introduced here, such as how to deal with high-branching and numeric attributes, missing values as well as how to prune. Taking the market's demand, bidding price and the capacity of bidding unit into consideration, the authors illustrate an example that indicates the procedure of calculation and some knowledge about load's rate and the attributes mentioned above. In this way, the data of bidding ability can be obtained. This knowledge is very useful in supporting the generating bidding unit to make decisions and the electric agency, PX and ISO to design an optimal trade project.
出处
《电力系统自动化》
EI
CSCD
北大核心
2002年第15期22-26,共5页
Automation of Electric Power Systems