«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

31

期数:

2021年03期

页码:

58-64

栏目:

大数据分析与挖掘

出版日期:

2021-03-10

文章信息/Info

Title:

Analysis of Role Selection in E-sports Based on Ordered Lasso-Logistic Model

文章编号:

1673-629X(2021)03-0058-07

作者:

王白云; 沈春根

上海理工大学 理学院,上海 200093

Author(s):

WANG Bai-yun; SHEN Chun-gen

School of Science,University of Shanghai for Science and Technology,Shanghai 200093,China

关键词:

机器学习; 有序 Lasso-Logistic 模型; 电竞; 预测; 先验信息

Keywords:

machine learning; ordered Lasso-Logistic model; e-sports; prediction; prior information

分类号:

TP39

DOI:

10. 3969 / j. issn. 1673-629X. 2021. 03. 010

摘要:

电竞行业近几年来迅速发展,这其中离不开机器学习在电竞中的分析和应用。职业选手能够熟悉电竞角色往往需要一定的时间和经验,采用机器学习对电竞角色进行分析,有利于选手对角色选择的考虑, 为职业选手的训练和比赛提供数据支持,体现预测模型的分析效果。 基于电竞结果通常为二分类数据,为了更好地利用先验信息,将自变量对因变量发生的不同重要性归结到 Lasso-Logistic 模型中形成有序 Lasso-Logistic 模型, 通过有序 Lasso-Logistic 模型分析电竞角色数据,对电竞结果进行预测。 根据结果显示,有序 Lasso-Logistic 模型预测效果显著, 将有序 Lasso-Logistic 模型预测结果与逻辑回归模型,Lasso-Logistic 模型,梯度增强决策树模型以及 SVM 模型的预测结果相比较,增加了先验信息的有序Lasso-Logistic 模型的预测效果明显比其他四个机器学习模型更佳,表明了有序 Lasso-Logistic 模型在机器学习中对分类数据相比经典分类模型具有更好的分析能力,突出自变量先验信息重要的作用,推动了机器学习在电子竞技行业的应用与发展。

Abstract:

The e-sports industry has developed rapidly in recent years,which cannot be separated from the analysis? ?and application of machine learning. Professional players can be familiar with the e-sports roles usually? ?need a certain amount of time and experience. Analyzing the e-sports roles by machine learning is cond-ucive to the player’s consideration of roles selection and provides data support for the training and compe-tition of profes-sional players,showing performance of predicted model. The e-sports results are usually bi-nary data. In order to make better use of prior information,the different importance of independent variables? ?to the occurrence dependent variables is combined with Lasso -Logistic model to form an ordered Lasso-Logistic model. We analyze the e-sports roles data by ordered Lasso-Logistic model and predict the comp-etition results. Accor-ding to the results,the ordered Lasso -Logistic model has a significant prediction effect. Taking the performa-nce of the ordered Lasso-Logistic model compared with the logistic regression model,? the Lasso-Logistic model, the Gradient Boosting Decision Tree model and the SVM model,we found that the prediction of ordered Lasso Logistic definitely better than the others,which suggests that the ordered model? ?is the better analytical skills for classified data in machine learning and importance of prior information of independent variables and promoted the application and development of machine learning in e-sports industry.

常用功能

更新日期/Last Update: 2020-03-10