Machine Learning and Data Mining for Sports Analytics: 9th International Workshop, MLSA 2022, Grenoble, France, September 19, 2022, Revised Selected Papers

Description

This book constitutes the refereed proceedings of the 9th International Workshop on Machine Learning and Data Mining for Sports Analytics, MLSA 2022, held in Grenoble, France, during September 19, 2022.  The 10 full papers included in this book were carefully reviewed and selected from 18 submissions. They were organized in topical sections as follows: Football, Racket sports, Cycling. Preface Organization Contents Football Towards Expected Counter - Using Comprehensible Features to Predict Counterattacks 1 Introduction 2 Framework for Understanding Complex Sequences 3 Definition of Sequences of Interest and Success Criteria 3.1 Rule-based Identification of Persistent Open-Play Turnovers 3.2 Definition of Success Criteria for Counterattacks 3.3 Emerging Dataset 4 Comprehensible Features for Prediction 4.1 Constructing Features from Domain-Specific Assumptions 4.2 Influence of Ball Loss Location for Feature Assessment 4.3 Prediction Capability of the Constructed Features 5 Model-based Test of Features 6 Conclusion References Shot Analysis in Different Levels of German Football Using Expected Goals 1 Introduction 2 Related Work 3 Methodology 3.1 Data 3.2 Statistical Analysis 3.3 Expected Goals Models 4 Results 4.1 Statistical Analysis 4.2 Expected Goals Models 5 Conclusions A Box plots of significantly different distributions References Analyzing Passing Sequences for the Prediction of Goal-Scoring Opportunities 1 Introduction 2 Problem Definition 3 Methodology 3.1 Tracking Data 3.2 Event Data 3.3 Data Alignment 3.4 Extraction of Goal Scoring Opportunities 3.5 Pitch Partitioning 3.6 Sequential Pattern Mining 4 Experimental Study 5 Style of Play for the Top-2 Teams 6 Conclusions and Future Work References Let's Penetrate the Defense: A Machine Learning Model for Prediction and Valuation of Penetrative Passes 1 Introduction 2 Related Work 3 Penetrative Pass Prediction and Valuation 3.1 Dataset and Preprocessing 3.2 Potential Penetrative Pass Situation 3.3 Penetrative Pass Label Generation 3.4 Penetrative Pass Decomposed Model 4 Experiments and Results 4.1 Best Performing Prediction Model 4.2 Does a Penetrative Pass Affect Goal Scoring or Conceding? 4.3 Teams' Penetrative Performance Analysis 4.4 Field Section Analysis: 5 Conclusion References Evaluation of Creating Scoring Opportunities for Teammates in Soccer via Trajectory Prediction*-12pt 1 Introduction 2 Proposed Framework 2.1 Potential Score Model in Modified OBSO 2.2 C-OBSO with Trajectory Prediction 3 Experiments 3.1 Dataset 3.2 Data Processing for Verification 3.3 Our Model Verification 3.4 C-OBSO Results 4 Related Work 5 Conclusion A Overview of our Method B Off-Ball Scoring Opportunity ch5Spearman18 C Variational Recurrent Neural Network ch5Chung15 D Graph Variational Recurrent Neural Network ch5Yeh2019 E Validation Results of Trajectory Prediction Model F C-OBSO and OBSO Results Without the Potential Score Model G Relationship Between Rating, C-OBSO, and Goal References Cost-Efficient and Bias-Robust Sports Player Tracking by Integrating GPS and Video 1 Introduction 2 Related Work 2.1 Optical Player Tracking 2.2 GPS-Based Player Tracking 2.3 GPS-OTS Integration Approach 3 Main Contributions 3.1 Anchor Starter Detection 3.2 Anchor Segment Detection 3.3 GPS-OTS Trajectory Matching per Anchor Segment 3.4 Initial Estimation of GPS Biases 3.5 Fine-Tuning GPS Biases 4 Experiments 4.1 Data Preparation 4.2 Implementation Detail 4.3 Model Evaluation 5 Conclusion and Future Work References Racket Sports Predicting Tennis Serve Directions with Machine Learning 1 Introduction 2 Related Work 3 Basic Information About Tennis Serves 4 Data 5 Feature Engineering 5.1 Outcome of Previous Points 5.2 Fatigue 5.3 Performance Anxiety 5.4 Other Features 6 Machine Learning 7 Discussion 8 Conclusion and Future Work References Discovering and Visualizing Tactics in a Table Tennis Game Based on Subgroup Discovery 1 Introduction 2 Methodology 2.1 Dataset 2.2 Tactics in Table Tennis 2.3 Mining Frequent and Discriminant Sequential Pattern 2.4 Summary of Assumptions 3 Results 3.1 Presentation of the Obtained Alternate Sequences 3.2 Visualization of the Tactics 4 Conclusion and Perspectives A Appendix References Cycling Athlete Monitoring in Professional Road Cycling Using Similarity Search on Time Series Data 1 Introduction 2 Related Work 3 Materials 3.1 Materials 3.2 Data Preprocessing 4 Methodology 4.1 Selection of Potential Matches 4.2 Taylor-made Approach 4.3 Dimensionality Reduction Approach 5 Results 5.1 Modeling Performance 5.2 Athlete Monitoring 6 Discussion 7 Conclusion References Author Index