| 基于类别比例因子和类内均分度的χ2统计改进 |
| |
| 引用本文： | 张瑜,张德贤.基于类别比例因子和类内均分度的χ2统计改进[J].电子科技,2010,23(12):70-72. |
| |
| 作者姓名： | 张瑜 张德贤 |
| |
| 作者单位： | (河南工业大学 信息科学与工程学院,河南 郑州〓450001) |
| |
| 基金项目： | 河南省创新性科技团队建设基金资助项目(094200510009) |
| |
| 摘 要： | 针对χ2统计特征选择方法的两大局限:对低文档频的特征选择不合理,以及过分强调那些在指定类低频出现,而在其他类中高频出现的特征项在该类中的权重。提出基于类别比例因子与类内均分度的χ2统计特征选择的改进方法。实验结果表明,改进方法的分类效果优于传统方法。
|
| 关 键 词： | 特征选择 &chi 2统计 类别比例因子 类内均分度 |
Improvement χ2 of Statistics Based on the Category Scale Factor and Average Distribution Inner Category |
| |
| Zhang Yu,Zhang Dexian.Improvement χ2 of Statistics Based on the Category Scale Factor and Average Distribution Inner Category[J].Electronic Science and Technology,2010,23(12):70-72. |
| |
| Authors: | Zhang Yu Zhang Dexian |
| |
| Affiliation: | (College of Information Science and Technology,Henan University of Technology,Zhengzhou 450001,China) |
| |
| Abstract: | The χ^2statistical method has two defects.One has reduced the weight of the low-frequency terms and the other has increased the weight of the characteristics in the designated class.This paper proposes an improved χ^2statistical approach based on the category scale factor and average distribution inner category.A contrastive experiment is carried out and the results show that improved χ^2statistics is superior to traditional statistical in feature selection. |
| |
| Keywords: | feature selection χ^2 statistics category scale factor average distribution inner category |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
| 点击此处可从《电子科技》浏览原始摘要信息 |
|
点击此处可从《电子科技》下载全文 |
