试析偏最小二乘回归模型对短跑成绩的预测  

Partial Least-squares Regression Model Based on Athlete Selection Index to Predict Sprint Performance

作　　者：陈显健[1] CHEN Xian-jian(Ningbo Polytechnic,Ningbo 315800,China)

机构地区：[1]宁波职业技术学院,浙江宁波315800

出　　处：《浙江工商职业技术学院学报》2021年第2期48-52,共5页Journal of Zhejiang Business Technology Institute

摘　　要：当前对人体短跑运动能力的评价主要采用定性研究,主观性较强,可运用一种多元数据分析方法即偏最小二乘回归分析方法(PLSR),对人体短跑运动能力的各影响因素进行定量分析,建立短跑运动成绩的预测模型。在随机选取了25名17-21岁大学校集训队短跑运动员和25名普通大学生后,测量受试者身高、腿长、小腿长+足高、比大小腿长、踝围、跟腱长、比跟腱长、比踝围、血色素、体重、肺活量、比肺活量、台阶指数、立定跳、反应时等15个指标作为解释变量并将样本随机的分为校正集和测试集,分别占样本集的三分之二和三分之一。运用SPSS 22.0软件进行相关性分析,对数据进行偏最小二乘回归分析(PLSR)建立预测模型,并进行三项PLSR建模实验,最终选取一个实验的模型作为预测集。研究发现,短跑运动成绩预测模型能合理有效地对短跑运动员的潜力进行定量分析,且精度合理,有望在运动选材领域得到推广应用。At present, the evaluation of human sprinting ability mainly adopts qualitative research, with strong subjectivity. A multivariate data analysis method, partial least squares regression analysis(PLSR), can be used to quantitatively analyze the influencing factors of human sprinting ability, and establish the prediction model of sprinting performance. The study randomly selected 25 sprinters and 25 ordinary college students, and used their15 indicators as explanatory variables. The samples were randomly divided into calibration set and test set,accounting for two-thirds and one-third of the sample set respectively. SPSS 22.0 software was used for correlation analysis, partial least squares regression(PLSR) was used to establish the prediction model, and three PLSR modeling experiments were carried out. Finally, an experimental model was selected as the prediction set.It is found that the prediction model of sprint performance can analyze the potential of sprinters reasonably and effectively, and the accuracy is reasonable, which is expected to be popularized and applied in the field of sports material selection.

关 键 词：偏最小二乘回归(PLSR) 运动选材 短跑成绩 

分 类 号：G80-059[文化科学—运动人体科学]