机构地区: 江西师范大学体育学院
出 处: 《中国体育科技》 2010年第6期138-141,共4页
摘 要: 偏最小二乘法(PLS)是一种集多元线性回归法(MLR)和主成分回归法(PCR)的基本功能于一体,对高维度数据具有强大的处理能力的模式识别方法,目前已在生物信息学、药学、社会科学等领域得到了广泛的应用,而在体育学领域,PLS的研究与应用相对缓慢,其功能还有待于更多的研究与开发。以参加第15届亚运会中短距离比赛的中国游泳队男运动员的核磁共振(NMR)数据为例,通过与SPSS软件中常用的主成分分析法(PCA)降维及线性判别分析(LDA)数据处理效果进行比较,阐述PLS分析的优越性,以及如何利用PLS进行降维、发现异常数据、分析自变量因子(各观测指标)的重要性程度和实现判别分析。 Partial least squares(PLS)analysis includes the basic function of multiple linear regression and principal component regression.It is a pattern recognition method with strong processing power for high-dimensional data.Currently,the partial least squares have been widely used in the field of bioinformatics,pharmacy and social sciences.However,in the field of sports science,the studies and application of partial least squares are lagged behind,and still need further studied and developed.Based on the NMR data from the Chinese middle-short distance male swimmers at the 15th Asian Games,this paper introduced the basic principle of PLS analysis,and how to use PLS to reduce dimension,found out the exceptional data,analyzed the importance of independent variables(observability index) and carried out the discriminant analysis.