| 一种基于特征加权的K Nearest Neighbor算法 |
| |
| 引用本文： | 桑应宾,刘琼荪.一种基于特征加权的K Nearest Neighbor算法[J].海南大学学报(自然科学版),2008,26(4):352-355. |
| |
| 作者姓名： | 桑应宾 刘琼荪 |
| |
| 作者单位： | 重庆大学数理学院,重庆,400044 |
| |
| 摘 要： | 传统的KNN算法一般采用欧式距离公式度量两样本间的距离．由于在实际样本数据集合中每一个属性对样本的贡献作用是不尽相同的，通常采用加权欧式距离公式．笔者提出一种计算权重的方法，即基于特征加权KNN算法．经实验证明，该算法与经典的赋权算法相比具有较好的分类效果．
|
| 关 键 词： | 特征权重 K近邻 交叉验证 |
A Weighting-based on Feature of KNN Algorithm |
| |
| SANG Ying-bin,LIU Qiong-sun.A Weighting-based on Feature of KNN Algorithm[J].Natural Science Journal of Hainan University,2008,26(4):352-355. |
| |
| Authors: | SANG Ying-bin LIU Qiong-sun |
| |
| Institution: | SANG Ying-bin,LIU Qiong-sun (College of Mathematics , Physics,Chongqing University,Chongqing 400044,China) |
| |
| Abstract: | The traditional method about KNN is generally used Euclidean distance formula to measure the distance between the two samples.However,every attribute has different function in actual samples;the weighted Euclidean distance is usually chosen.This paper presents a calculation of weight that is weighted based on the characteristics of KNN algorithm.The results of the experiments on artificial datasets show that this algorithm can improve the accuracy of classification. |
| |
| Keywords: | feature weighted K nearest neighbor cross validation |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
