Design and Implementation of NBA Playoff Prediction Method Based on ELO Algorithm and Graph Database

With the globalization of NBA, all eyes on the NBA playoffs are around the world. Ones celebrate the winning of their team which they like. Especially, NBA fans keep on predicting the playoffs game results. However, prediction of winning probability of teams in NBA playoffs is challenging. In order to meet the challenges, we proposed a method using ELO algorithm for prediction and leveraging Graph Database, Neo4j, for implementation. Experiment results show that, the design and implementation of the prediction system could work to some degree.

基于ELO体系的学科竞赛排名模型优化研究

随着越来越多的高校学生投入到学科竞赛中,急需对参与竞赛的选手进行公开、透明且有效的评分和选拔。先将ELO评分系统从双人博弈拓展至多人,考虑到多人ELO评分系统收敛速度慢,引入定位赛的机制,通过定位系数使选手分数快速收敛,保证可靠性,并通过排名优化、机遇优化解决比赛排名和选手实力严重不符的问题,采用膨胀优化解决ELO评分系统总分数不断膨胀的问题。针对模拟数据和团队数据的实验结果表明优化的ELO评分系统在实际运用中能较为精准的反应选手的水平,在极端情况下的稳定性较好。