基于类别信息熵加权的MKNN算法

MKNNAlgorithm Based on the Weight of Category's Entropy

针对MKNN算法对类属性数据处理简单的问题,引入信息熵作为处理类属性数据的相似性度量,进而引入类别信息熵的概念。对同一类型的类属性数据根据其类别信息熵权重的大小,把数据集的记录进行分类进而得到测试结果。实验结果验证了该算法的有效性。

Since the process of the mutual k-nearest neighbor(MKNN) dealing with nominal data is simple, we introduce the entropy to deal with the similarity measure of the nominal data, and then the concept of Category's entropy is introduced. We can obtain the entropy weight of the same type of nominal data, and then get experimental results through the classification of the data set. The experimental results demonstrate the effectiveness of the proposed algorithm.