基于秩和检验与SVM的基因特征选取与分类方法

研究了如何提取基因图谱信息的方法问题,提出一种新的基因"标签"提取方法.该方法采用秩和检验法选取基因"标签",得到不同显著性水平下基因"标签"的数目,然后将秩和检验法和分类信息指数相结合,在与基于SVM建立的分类模型比较后,提取最合适的基因"标签",再结合分类信息指数重复数次对这些基因"标签"选取主基因,最后用SVM检验主基因选取及分类的准确率,发现准确率有所提高,说明该方法的有效性.

Extracting Information and Classifying Method on Gene Based on Ranksum-Test and SVM

This paper studies the methods of how to extract information on gene map, A new extraction method is proposed based on genetic ＂signatures＂, the rank and test methods are used to select genetic ＂signature＂. The number of genetic ＂signatures＂ is obtained under different significance level, and then combines the methods of ranksum-test and the Classified Index informarion, through comparing with the classification model; we can extract the most appropriate genetic ＂signatures＂. The repeated Index times of classified information is introduced to selected major gene from those genetic ＂signatures＂. Finally, through the rate of accuracy of the gene selection and classification, we obtained that the rate of accuracy is better than before, which indicate that the effectiveness of this method.