Abstract: An improved KNN algorithm based on contribution of attribute value and average similarity is proposed，which is different from the traditional K-nearest neighbor algorithm.This paper not only considers the average similarity between the test samples and the similar samples but also considers the number of the similar samples in different categories.And finally considers the contribution of the same attribute value to the class label.The experiments is performed on mushroom data sets.The experimental results show that the approach has higher accuracy than the traditional K-nearest neighbor algorithm.

Key words: classification, K-Nearest Neighbor algorithm（KNN）, similarity

摘要：

与传统的K-近邻算法不同，提出了一种结合属性值贡献度与平均相似度的KNN改进算法。首先考虑测试样本与相似样本点间的平均相似度，其次考虑不同类别中的相似样本点的个数，最后还考虑与相似样本相同的属性值对类别的贡献度。在蘑菇数据集上进行实验结果表明，改进后的KNN分类算法的准确率比传统的K-近邻分类算法的准确率更高。

关键词: 分类, K-近邻算法, 相似度