基于多元逐步回归的RoboCup边界球研究  

Research on RoboCup boundary ball based on multiple linear regression model

作　　者：余培 Yu Pei(College of Computer,Anhui University of Technology,Ma’anshan 243000,China)

机构地区：[1]安徽工业大学计算机学院,安徽马鞍山243000

出　　处：《六盘水师范学院学报》2020年第3期70-74,共5页Journal of Liupanshui Normal University

基　　金：2019年度国家级大学生创新创业训练项目“基于Python的RoboCup仿真足球赛边界球技术的研究”(201910360080)。

摘　　要：边界球是RoboCup前人研究中较少涉及的一个研究点,为提升球队胜率,探究边界球分布和比分之间的关系。采用数据挖掘的方式,使用Python语言解析日志文件,将球场划分为四个区域,将左右队伍在各区域的边界球数之差作为四个解释变量,将比分差作为因变量,采用逐步回归法筛选显著的解释变量,搭建多元线性回归模型,得到使得球队胜率较高的边界球分布模型,最后结合实际比赛中边界球和铲球的相关性,得到应加强[-52.5,-26.25)区域界外铲球和[-26.25,0)区域界内铲球的结论。Boundary ball is a research point seldom involved in RoboCup’s previous studies.In order to im⁃prove the team’s victory rate,the relationship between the distribution of boundary balls and the score is ex⁃plored.Using data mining method and Python language to analyze log files,the stadium is divided into four re⁃gions,the difference between the number of boundary balls of left and right teams in each region is taken as four explanatory variables,the score difference is taken as dependent variable,the significant explanatory variables are screened by stepwise regression method,a multiple linear regression model is built,and a boundary ball dis⁃tribution model with high team victory rate is obtained.Finally,combining the correlation between boundary balls and tackle balls in actual matches,the conclusion that[-52.5,-26.25)area out-of-bounds tackle balls and[-26.25,0)area in-bound tackle balls should be strengthened is obtained.

关 键 词：机器人世界杯 数据挖掘 日志挖掘 回归分析 逐步回归 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]