| 基于模糊粗糙最近邻算法的不平衡数据分类 |
| |
| 引用本文： | 丁兴波.基于模糊粗糙最近邻算法的不平衡数据分类[J].教育技术导刊,2009,19(11):37-41. |
| |
| 作者姓名： | 丁兴波 |
| |
| 作者单位： | 南京信息职业技术学院 人工智能学院，江苏 南京 210023 |
| |
| 摘 要： | 为了提升不平衡数据中少数类的分类精度，利用SMOTE采样方法对数据集进行平衡化预处理；为了减轻样本重新合成过程中产生的类重叠和噪声对分类精度的影响，选择模糊粗糙最近邻算法（FRNN）作为分类器。在14个不平衡数据集上进行的仿真实验表明，该方法具有较好的分类表现，F值和G值最高分别可达0.965、0.932，是一种适用于不平衡率偏高数据集的分类方法。
|
| 关 键 词： | 不平衡数据 分类器 SMOTE 模糊粗糙最近邻算法 |
| 收稿时间： | 2020-07-09 |
A Classification Method for Imbalanced Data Based on Fuzzy Rough Nearest Neighbor |
| |
| ZHANG Chun-mei.A Classification Method for Imbalanced Data Based on Fuzzy Rough Nearest Neighbor[J].Introduction of Educational Technology,2009,19(11):37-41. |
| |
| Authors: | ZHANG Chun-mei |
| |
| Institution: | Institute of Artificial Intelligence， Nanjing Vocational College of Information Technology， Nanjing 210023，China |
| |
| Abstract: | In order to improve the classification accuracy of the minority classes in imbalanced data， the paper employs synthetic minority over - sampling technique（SMOTE） to balance data set firstly. Considering that the process of sample re-synthesis always leads to some noises such as class overlapping， fuzzy rough neareswast neighbor algorithm （FRNN） is selected as the classifier to alleviate the effect of noise. Classification experiment conducted on 14 unbalanced data sets shows that the proposed method performs well， and the F value and G value can reach 0.965 and 0.932 respectively. It reveals that the proposed method is suitable for the classification on data sets with high imbalance rate. |
| |
| Keywords: | imbalanced data classifier SMOTE fuzzy rough nearest neighbor algorithm |
| 本文献已被 维普 等数据库收录！ |
| 点击此处可从《教育技术导刊》浏览原始摘要信息 |
| 点击此处可从《教育技术导刊》下载免费的PDF全文 |
