一种基于属性加权的决策树算法

A Decision Tree Algorithm Based On Attribute Weight

ID3算法和C4.5算法是简单而有效的决策树分类算法,但其应用于复杂决策问题上存在准确性差的问题。本文提出了一种新的基于属性加权决策树算法,基于粗集理论提出通过属性对决策影响程度的不同进行加权来构建决策树,提高了决策结果准确性。通过属性加权标记属性的重要性,权值可以从训练数据中学习得到。实验结果表明,算法明显提高了决策结果的准确率。

ID3 algorithm and CA. 5 algorithm are all simple and productive decision tree classification algorithms, but they have low accuracy on complex decision problems such as exploratory wells decision support system. In the paper, a decision tree algorithm based on attribute weight is proposed to improve accuracy of the decision. Determining Weights is according to different influence degrees of decision of different attributes. Weights deriving from training data mark the importance of attributes. Experiment shows that the algorithm is efficient on improvement of accuracy.