连续属性离散化的Imp-Chi2算法

Imp-Chi2 Algorithm for Discretization of Real Value Attributes

连续属性离散化是机器学习和数据挖掘领域中的一个重要问题,离散化是否合理决定着表达和提取相关信息的准确性。经过研究Chi2系列算法,提出一种新的基于属性重要性的连续属性离散化方法——Imp-Chi2算法,该算法依据属性重要性程度对属性离散化的顺序进行了合理的调整,能够更准确地对连续属性进行离散化。文章通过C4.5和支持向量机分别对离散化后的结果进行了实验,在实验过程中,提出一种训练集类比例抽取方法,避免了训练集随机抽取的不均匀性。实验结果证明了所提算法的有效性。

Discretization is an effective technique to deal with continuous attributes for machine learning and data mining. Reasonability of a discretization process is determined by the accuracy of expression and extraction for informations. By analyzing a series of Chi2 algorithm, a new algorithm called Imp-Chi2 algorithm is proposed, which is based on attribute significance. The algorithm reasonably adjusts the sequence of disretization for attributes according to the level of attribute significance, and exactly discretes the real value attributes. The experiments are performed respectively with the results of discreted data by using C4.5 and SVM. In the process of the experiments, a selection method of training set according to class proportion is presented. The method overcomes the bad-distributed situation for random selection of training set. Experimental results show that the presented algorithm is effective.