Machine Learning and Data Mining for Sports Analytics: 10th International Workshop, MLSA 2023, Turin, Italy, September 18, 2023, Revised Selected Papers ... Computer and Information Science Book 2035)

This book constitutes the refereed proceedings of the 10th International Workshop on Machine Learning and Data Mining for Sports Analytics, MLSA 2023, held in Turin, Italy, in September 2023.The 16 full papers included in this book were carefully reviewed and selected from 31 submissions. They were organized in topical sections as follows: Football/Soccer, Basketball, Other team sports, Individual sports. Preface Organization Contents Keynote Sports Data Analytics: An Art and a Science 1 Definition of Sports and Its Consequencies for Sports Data Analytics 1.1 It's All About Skill 1.2 What Contributes to Complexity? 2 How Do We Develop Meaningful Skill Metrics in Complex Sports? 2.1 How We Deal with Complexity? 2.2 Skills on the Individual Level 2.3 Skills on the Team Level 3 Conclusion References Football/Soccer ETSY: A Rule-Based Approach to Event and Tracking Data SYnchronization 1 Introduction 2 Problem Statement and Existing Approaches 3 ETSY 3.1 Preprocessing 3.2 Synchronization 3.3 Action-Specific Filters 4 Evaluation 4.1 Experimental Setup 4.2 Experimental Comparison 4.3 Missed Events Analysis 4.4 Example Synchronization 4.5 Conceptual Comparison 5 Conclusion References Masked Autoencoder Pretraining for Event Classification in Elite Soccer 1 Introduction 2 Preliminaries and Problem Formulation 3 Trajectory Masked Autoencoder 3.1 Masked Autoencoder for Multiagent Trajectories 3.2 Factorized Transformer Encoder (FTE) 4 Event Classification in Elite Soccer 5 Conclusion References Quantification of Turnover Danger with xCounter 1 Introduction 2 Dataset 3 Application of the Framework for Understanding Complex Sequences 3.1 Sequences of Interest 3.2 Success Criteria 3.3 Comprehensible Features 3.4 Feature-Specific Prediction Capability 3.5 Predictive Model 3.6 Design Choices 4 Results and Discussion 4.1 Prediction Capability of Features 4.2 Predictive Models for Expected Counter 5 Related Work 6 Application 7 Limitations and Conclusion A Appendix References Pass Receiver and Outcome Prediction in Soccer Using Temporal Graph Networks 1 Introduction 2 Related Work 3 Decomposing the Pass Success Probability 4 Constructing Temporal Graph Networks 4.1 Model Definition 4.2 TGN Architecture 4.3 Training RSP and RPP Models 5 Experiments 5.1 Dataset 5.2 Evaluating Model Performance 5.3 Pass Difficulty in Different Areas of the Pitch 6 Conclusion References Field Depth Matters: Comparing the Valuation of Passes in Football 1 Introduction 2 Methodology 2.1 Event Data 2.2 VAEP 2.3 Groups and Tests 3 Results 3.1 Comparing Same Third Groups 3.2 Comparing 1-G and 3-NG 3.3 Notable Cases 4 Conclusion and Future Work 4.1 Conclusion 4.2 Future Work References Basketball Momentum Matters: Investigating High-Pressure Situations in the NBA Through Scoring Probability 1 Introduction 2 Pre-game Pressure 3 Identifying High-Pressure Scenarios 4 Evaluation of Players' Performance in High-Pressure Situations 5 Conclusion and Future Work References Are Sports Awards About Sports? Using AI to Find the Answer 1 Introduction 2 Background 3 Experimental Overview 3.1 Data Processing 3.2 Implementation Details 4 Results and Analysis 4.1 Predicting the Most Valuable Player 5 Conclusion and Future Work References The Big Three: A Practical Framework for Designing Decision Support Systems in Sports and an Application for Basketball 1 Introduction 2 Literature Overview 3 The Big Three: Model, Explanation and Interactivity 4 BasketXplainer: A DSS for Basketball 5 Results and Analysis 5.1 Model 5.2 Explainability and Interactivity 6 Conclusion and Future Work 6.1 Research Opportunities References Other Team Sports What Data Should Be Collected for a Good Handball Expected Goal model? 1 Introduction 2 State of the Art 2.1 The xG Metric 2.2 Use of xG 3 Handball Data 3.1 Data Acquisition 3.2 Qatar 2015 Data 3.3 Dataset Enrichment 4 Method for Calculating xG 4.1 Sample Calculation 4.2 Model Calculation 4.3 Wrapper: Attribute Selection Approach 5 Results 5.1 Sample Calculation 5.2 Wrapper Approach 6 Conclusion References Identifying Player Roles in Ice Hockey 1 Introduction 2 Data 3 Method 3.1 Preprocessing 3.2 Principal Component Analysis 3.3 Clustering 4 Results 4.1 Principal Component Analysis 4.2 Fuzzy Clustering 5 Comparing Player Salary 6 Team Composition 7 Conclusion References Elite Rugby League Players' Signature Movement Patterns and Position Prediction 1 Introduction 2 Method 2.1 GPS Data, Collection and Processing 2.2 Signature Movement Patterns Mining 2.3 Playing Position Prediction and Feature Contribution 3 Results and Discussion 3.1 Signature Movement Patterns 3.2 Playing Position Prediction 4 Conclusion and Future Works References Boat Speed Prediction in SailGP 1 Introduction 2 Background 3 Data Set 4 Boat Speed Analysis 4.1 Unveiling the Key Component 4.2 Random Forest Model 4.3 Gradient Boosted Model 4.4 Benchmark 4.5 Meta Model 5 Conclusion References Individual Sports Exploring Table Tennis Analytics: Domination, Expected Score and Shot Diversity 1 Introduction 2 Domination Analysis in Table Tennis 3 Expected Score (XScore) in Table Tennis 4 Shots Diversity in Table Tennis 5 Discussion, Limits and Perspectives A Appendix A.1 Definition of the Winning Probability Pa,b A.2 Definition of the Three Factors of Physical Domination A.3 Definition of the Three Factors of Mental Domination References Performance Measurement 2.0: Towards a Data-Driven Cyclist Specialization Evaluation 1 Introduction 2 Literature Review 3 Methodology 3.1 Data 3.2 Clustering 3.3 Performance Ranking 4 Results 5 Conclusion References Exploiting Clustering for Sports Data Analysis: A Study of Public and Real-World Datasets 1 Introduction 2 Related Work 3 Clustering Comparison 4 Evaluation on Real-World Data Set 5 Conclusion References Author Index