摘要
提出用C4.5决策树方法解决负荷预测的样本多样性问题,并进行短期负荷预测。通过计算信息增益找出决策树的最佳生成方案,对连续属性计算其熵值找出最佳分段点进行离散化,阐述了规则的生成及其在短期电力负荷预测中的应用方法,算例结果表明,计算精度较高。
This paper presents the decision tree method to forecast the electrical load with diverse samples. Entropy and information gain is calculated to get the best decision tree. Entropy is also used to disperse continuous data attributes and get the best splitting point. The paper describes the way of generating decision tree hales which is applied to short-term load forecasting. Calculation result by the method on a real power grid is proves the applicability of the presented method.
出处
《华中电力》
2009年第1期15-18,共4页
Central China Electric Power
关键词
决策树
负荷预测
熵
信息增益
离散化
数据挖掘
decision tree
load forecasting
entropy
information gain
decision tree dispersion
highly precise, which data mining