[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-127323-en":3,"doc-seo-127323-105":30,"detail-sidebar-cat-0-en-105":92},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},127323,962085570644,"Evangeline","https://ap-avatar.wpscdn.com/davatar_994ba38a5ba835b3df7d355c54d3ed8d",8,"Research & Report","A Graph Machine Learning Approach to Estimating Scoring Probabilities and Drive Outcomes in Gridiron","Machine learning models for score outcome probability and drive outcome probability are introduced and tested using play-by-play data for the NFL and CFL. Graph Convolutional Network (GCN) approaches are emphasized, incorporating both per-play features and information from immediately preceding plays when producing scoring predictions. A GCN with drive information, a GCN without drive information, and two non-graph baselines are compared, with further fine-tuning experiments performed. Results show graph models are outperformed by well-tuned MLPs, while removing previous-play context yields no meaningful accuracy or calibration change.","A Graph Machine Learning Approach to Estimating Scoring Probabilities and Drive Outcomes in Gridiron  \nFootball  \nBaxter Madore  \nSubmitted in partial fulfilment  \nof the requirements for the degree of Bachelor of Science, Honours Computing Science  \nat  \nSaint Mary’s University  \nHalifax, Nova Scotia  \n➞ Copyright Baxter Madore, 2025  \nApproved by: Dr. Jiju Poovvancheri  \nSupervisor  \nDr. Paul Muir  \nReader  \nDate: May 1, 2025  \nAbstract  \nMachine learning models for score outcome probability and drive outcome probability for the National Football League (NFL) and the Canadian Football League (CFL) are introduced and tested on play-by-play data. Of particular interest are Graph Convolutional Network (GCN) models, which in addition to training on features of individual plays also use information about the immediately preceding plays when generating their scoring predictions. A GCN model using drive information, a GCN model without drive information, and two non-graph models are compared, with additional testing to fine-tune models. The GCN model does not significantly change in accuracy or calibration when not using information about previous plays in the series, but both types of graph models are outperformed by well-tuned MLPs.  \nDate: May 1, 2025  \nAcknowledgments  \nI would first like to thank my supervisor, Jiju Poovvancheri, for his valuable insight into machine learning methods, an area I knew little about when I started this project eight months ago, and for taking me into his research group this year. Thanks to Graphics and Spatial Computing group members Prachi Kudeshia and Abdiaziz Muse for debugging help, especially in regards to the Compute Canada servers. I would also like to thank Paul Muir, for first introducing me to what academic research looks like, helping me develop my scientific skillset, and convincing me to switch into the Honours program.  \nTo everyone who I have had the pleasure of sharing the McNally North office with over the last two years, thank you for the conversation and sense of community.  \nContents  \nList of Tables iv  \nList of Figures v  \n1 Introduction 1  \n1.1 Introduction to Football ......................... 1  \n1.1.1 Scoring ............................... 2  \n1.1.2 Game Flow ............................ 3  \n1.1.3 Differences Between Canadian and American Football ..... 6  \n1.2 Objective ................................. 7  \n1.2.1 Drive Outcomes .......................... 9  \n1.3 Introduction to Football Models ..................... 10  \n1.4 Organization ............................... 11  \n2 Literature Review 13  \n2.1 History of Score Modelling in Football ................. 13  \n2.2 Graphs and GNNs in Sports Analytics ................. 18  \n3 Method 20  \n3.1 Data Acquisition and Treatment ..................... 20  \n3.1.1 Preprocessing ........................... 21  \n3.1.2 Data Splitting ........................... 23  \n3.1.3 Drive Outcomes .......................... 26  \n3.2 Multi-Layer Perceptron (MLP) Models ................. 27  \n3.2.1 Hyperparameters ......................... 32  \n3.3 Graph Neural Network (GNN) Models ................. 32  \n3.3.1 Why Graphs? ........................... 32  \n3.3.2 Creating the Graphs ....................... 35  \n3.3.3 GCN Model Explanation ..................... 36  \n3.3.4 Splitting the Graphs ....................... 40  \n3.3.5 GCN Model Training ....................... 41  \n3.4 xgboost .................................. 43  \n4 Results 44  \n4.1 Environment ................................ 44  \n4.2 Calibration ................................ 44  \n4.3 Models ................................... 48  \n4.3.1 MLP ................................ 48  \n4.3.2 xgboost .............................. 50  \n4.3.3 GCN with Temporal and Spatial Edges ............. 52  \n4.3.4 GCN without Temporal Edges .................. 53  \n4.3.5 Table of Results .......................... 54  \n5 Conclusion and Future Work 55  \n5.1 Conclusion ................................. 56  \n","cbCaic1xAke5ZdHc","https://ap.wps.com/l/cbCaic1xAke5ZdHc","pdf",2486826,1,74,"English","en",105,"# Contents\n## Introduction\n## Literature Review\n## Method\n## Results\n## Conclusion and Future Work","[{\"question\":\"What prediction tasks does the document address in gridiron football analytics?\",\"answer\":\"It models score outcome probability and drive outcome probability using play-by-play data for the NFL and CFL.\"},{\"question\":\"How do the graph-based models differ regarding the use of prior plays?\",\"answer\":\"One GCN uses drive information (immediately preceding plays), while another GCN omits that prior-play context.\"},{\"question\":\"Which model type performed best after fine-tuning?\",\"answer\":\"Well-tuned MLPs outperformed both graph model variants in accuracy and calibration.\"}]","A Graph Machine Learning Approach to Estimating Scoring Probabilities and Drive Outcomes in Gridiron | PDF",1785938291,186,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":87,"head_meta":89,"extra_data":91,"updated_unix":28},"a-graph-machine-learning-approach-to-estimating-scoring-probabilities-and-drive-outcomes-in-gridiron","",{"@graph":36,"@context":86},[37,54,69],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/a-graph-machine-learning-approach-to-estimating-scoring-probabilities-and-drive-outcomes-in-gridiron/127323/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-22","2026-08-05",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"What prediction tasks does the document address in gridiron football analytics?","Question",{"text":76,"@type":77},"It models score outcome probability and drive outcome probability using play-by-play data for the NFL and CFL.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How do the graph-based models differ regarding the use of prior plays?",{"text":81,"@type":77},"One GCN uses drive information (immediately preceding plays), while another GCN omits that prior-play context.",{"name":83,"@type":74,"acceptedAnswer":84},"Which model type performed best after fine-tuning?",{"text":85,"@type":77},"Well-tuned MLPs outperformed both graph model variants in accuracy and calibration.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,124,129,132,136],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":46,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":46,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":46,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":46,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":46,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":46,"category_name":138,"show_sort_weight":107,"slug":139},19,"General","general"]