基于独立成分分析和随机森林判别法的Microarray分析及在分子生物学中的应用

Microarray analysis method based on independent component analysis and random forests discriminant

提出基于独立成分分析(ICA)和随机森林判别的Microarray分析方法。该方法先采用独立成分分析获取高阶统计信息,提取Microarray数据特征,达到降维的目的。再应用提取的特征,采用随机森林判别法对样本进行分类。数值分析结果表明,提取5个特征就可以使袋外样本OOB(out of bag)的分类错误率达到7.89%。该方法有效地降低了特征空间维数,具有较高的正确识别率,提高了算法的鲁棒性和灵活性。

In this paper, a microarray analysis method based on independent component analysis and random forests discriminant is provided. In this method the independent component analysis is used to obtain high order statistic information and extract features of microarray in order to reduce the dimension of the feature space. These features extracted were used to classify the samples of out - of -bag （OOB） by random forests discriminant. Numerical simulation shows that the classification error rates of OOB can be up to 7. 89% only by extracted five features. The method can reduce the dimension of feature space effectively and has higher correct classification rate. The results show that it improves robustness and flexibility of algorithms.