In an era where technology plays a pivotal role in achieving excellence, the University of Nevada Las Vegas (UNLV) has embraced the potential of data analytics and machine learning (ML) to enhance their athletic programs. Under the leadership of their Chief Information Officer, Sarah Thompson, a dedicated task force known as the Data Innovation Lab was established to tackle challenges across various departments and promote innovative technological solutions. One of the key areas identified for improvement was the analytics of sports performance through machine learning.
Guided by insights from professional sports, Jason Lee, the interim CIO, and the UNLV Athletics Department began to explore how data could revolutionize their approach to football coaching. To spearhead this initiative, they recruited David Morgan as the Director of Sports Analytics and Technology. Since his arrival, David has been instrumental in training the coaching staff to leverage data for game preparation.
David explains, “Every football game can be analyzed as a series of plays, where each play is influenced by various factors, including player positions, field conditions, and situational context. To maximize the likelihood of a win, coaches must enhance the chances of success for each individual play. I allow the insights from data to guide our strategies.”
However, extracting meaningful data insights is a complex task due to the multitude of variables involved in play-calling and the sheer volume of data available. Traditionally, coaching staff would manually create game planning sheets based on heuristics related to down, distance, and field position, often relying on a limited set of features from the datasets to determine optimal plays. This process is not only time-consuming but also challenging to scale.
To tackle these issues, David, the Data Innovation Lab, and the Amazon Machine Learning (ML) Solutions Lab collaborated to develop an ML-driven solution. Over several months, they created an ML model capable of generating weekly game planning sheets for coaches based on an extensive and ever-expanding dataset.
With the ML model in place, coaching staff can now produce updated game planning sheets each week, enhancing their strategic insights and saving valuable preparation time. Concurrently, Amazon Web Services (AWS) is conducting ML training sessions with the Data Innovation Lab, enabling them to continually refine the model and improve its predictive accuracy.
The approach taken by the team transformed the data into a supervised ML problem, where the outcomes of plays were categorized as successful or unsuccessful. They prepared the dataset by employing techniques such as one-hot encoding for categorical variables and balancing the dataset through upsampling. Utilizing Amazon SageMaker, they constructed an XGBoost model that achieved a weighted accuracy of 65.2 percent. During model training, Hyper-parameter Optimization (HPO) was leveraged to enhance performance further.
By analyzing feature importance scores from the XGBoost model, they identified relevant features for the new game planning sheets, which included opponent-specific insights suggested by the model. These were presented to the UNLV football coaches for feedback. Throughout the season, coaches will employ these templates to generate the most effective play types for various situations, executing a simple AWS Lambda function on the latest dataset. The XGBoost model will be reevaluated every few weeks to incorporate any new significant features into the game planning templates.
David and the UNLV coaching staff acknowledge that changing established coaching practices can be challenging, but they believe this evolution is crucial for the football program’s future. As David puts it, “We are proving to the coaching staff that there is much more to game preparation than just watching film. The data can narrate a compelling story for us.”
For more insights, check out this other blog post that elaborates on similar innovations. Also, for a deeper understanding of this topic, visit this link, as they are an authority in this field. Additionally, this is an excellent resource for newcomers seeking guidance in the realm of Amazon services: Reddit Resource.
Leave a Reply