Novel method for ranking batsmen in Indian Premier League

作     者：M.K.Manju Abin Oommen Philip 

作者机构：Department of Computer Science&EngineeringToch Institute of Science&TechnologyErnakulamKerala682313India Department of CSE-Cyber SecurityIndian Institute of Information TechnologyKottayam686635India 

出 版 物：《Data Science and Management》 (数据科学与管理（英文）)

年 卷 期：2023年第6卷第3期

页      面：158-173页

核心收录：

学科分类：08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Each feature used to evaluate the performance of the batsman was examined using supervised machine learning algorithms. For a detailed evaluation of the effectiveness of each feature contributing to the player performance index of the batsman  we incorporated different supervised machine learning models  namely  (1) function-based models  including logistic regression and support vector machine  and (2) decision tree-based classifiers  including RF  XGB classifier  CART  and Bayesian network models  such as Naive Bayes. These algorithms provide importance scores by reducing the criteria for selecting the split points based on the Gini index.This study proposes a novel framework that ranks batsmen in the IPL based on individual performances  areas of expertise  and quantifying players in their team collaboration. To the best of our knowledge  this is the first study and implementation of an all-round IPL ranking of these factors. Ranking the T20 batsmen for seasons (2008–2022) focused on three major steps: First  create a new performance index based on the primary statistics available in the dataset. The contribution of each feature to the ranking index was identified based on the feature importance values calculated using the decision-tree-based classifiers. The defined features were tested using function-based models (logistic regression and SVM) and decision tree-based classifiers (RF  XGB classifier  CART  and Bayesian network models such as NB). Second  cluster players based on the k-means clustering algorithm to identify the best players according to their roles using the newly created performance index. Role-based and team interaction-based evaluation performance indices were obtained using unsupervised learning algorithms and social network analysis approaches. Based on the newly created performance index  the k-means clustering algorithm was used to cluster players and identify the best players according to their roles. Third  measure the team interaction behavior of players based on the network theory approach in social network analysis using the network centrality measures of betweenness centrality  closeness centrality  and in-strength of the node. The networks for the players'’ batting partnership and the individual performance against each team were evaluated to assess the players’ team collaboration with other players. Players were ranked after each phase of the analysis by applying a suitable formula to the normalized performance index values of the players. The results show that the proposed method captures all possible aspects of batting strength based on individual performance  role  and team interaction  which is effective and offers a new framework for ranking players in any team sport 

主　　题：Indian Premier League Social network analysis Artificial intelligence Machine learning 

摘      要：Sports analytics have benefited immensely from the growth and popularity of artificial intelligence and machine *** techniques enable sports analysts to evaluate player performance more effectively.A literature review of player performance evaluation methods shows the need to develop a new performance evaluation index for Twenty20(T20)cricket.A novel framework was proposed to evaluate batsman strength based on individual performance,role in the team,and team ***,proposed ranking systems are derived from static networks,that is,the aggregation of game results over ***,the scores of the players(or teams)fluctuate over ***,defeating a renowned player during peak performance is more rewarding than defeating the same player during other *** account for this,we propose a new method and apply it to the T20 format Indian Premier *** method serves three main purposes:First,it creates a new performance index for players to rank them more accurately and ***,the players are clustered based on their *** the third phase,a social network analysis approach is applied to visualize and analyze crickets as a network to gain better insights into players’team *** novel approach is a helpful index for sports coaches,analysts,cricket fans,and managers to evaluate player performance and rank for future aspects.