基于中值PCA和加权PCA数据分类的研究

在分析了传统主成分分析（PCA）方法的原理和实现方法上,提出了基于中值的主成分分析新方法（MPCA）.另外,针对多类高维数据分类问题,较深入地研究了权函数对分类问题的影响,对传统PCA模型进行加权处理得到加权主成分分析（WPCA）.实验结果表明,MPCA比传统PCA具有较好的分类效果,不同权函数对数据的分类结果影响较大,且WPCA比传统PCA在分类效果上有明显的优势.

Studies on data classification based on median PCA and weighted PCA

A new median-based principal component analysis method （MPCA） is proposed under based on the traditional principal component analysis （PCA） theory and implementing method are introduced in this paper. In addition, the effect of weight function on the classification problem is deeply researched for multi-class high-dimension data classification problem. And, weighted principal component analysis （WPCA） is obtained by applying the weighting method for PCA model. The experiment results show that the MPCA has got better result than the traditional PCA on the classification problem, the influence of different weight functions is great on data classification result, and WPCA has obvious more advantages than the traditional PCA on classification results.