NBA季后赛成绩分析及预测:Logistic和Bayes模型

Analysis and Forecast of Playoff Teams Promotions with Logistic and Bayes Model

目的运用NBA常规赛数据统计分析季后赛球队,分析影响球队晋级的因素,预测判别球队晋级情况。方法因子分析对球队各项统计指标降维综合,提出了新的球队综合实力计算方法,根据NBA官方公式得到球员得分,结合主教练与主客场因素,建立Logistic模型与Bayes判别模型。结果新的球队综合实力计算方法优于官方公式;教练的执教能力对球队的晋级贡献最大;具有主场优势的球队只要获胜概率达到0.38则认为晋级;Logistic模型与Bayes判别模型判别正确率均较高。结论根据各指标建立的模型适用于所有季后赛球队,可用于预测季后赛晋级情况。

Objective By applying the 2004-2005,2005-2006,2006-2007 regular season statistics to analyze and find the factors of playoff teams＇ promotion.Methods A new calculating teams＇ scores method is developed based on factor analysis,and Logistic Regression Model and Bayes Discriminant Model are established.Results The coaches＇ capacity and experience is the first factor.Teams having home court advantage would likely to win the series on condition that its win probability is more than 0.38.The accuracy rates of discrimination by Logistic Regression Model and Bayes Discriminant Model are relatively high.Conclusions The models are applicable for all the playoff teams and to the forecast of promotion.