Inspiration

Attendance at sporting events is one of the strongest ways fans can show their support for their favorite team. Sports organizations can better assess the fan demand at a given sporting event and address their needs accordingly with greater insight into how popular one of their upcoming events is.

What it does

With Arena Insights, you can optimize your game day planning and preparation. Just find your organization, check your upcoming game schedule, and then choose a game for a prediction on how full your venue will be.

How we built it

After compiling relevant game and attendance data from previous NBA regular season games from online databases such as basketballreference.com and storing them using Amazon S3, we hosted and trained our machine learning model using Scikit-learn in Amazon SageMaker. We additionally stored pertinent information regarding future regular season games in S3. When the user selects an upcoming game, our model accesses its relevant data and gives a prediction on the percentage of arena seats that will be filled.

Challenges we ran into

1) Determining which factors to include in our machine learning model and finding sufficient data for them. Attendance at sports games can be influenced by a large number of different factors. This posed a significant challenge for us in finding the most important factors to consider. Additionally, some of the factors that we considered were difficult to find quantifiable data for.

Accomplishments that we're proud of

1) We were able to achieve 84.42% accuracy on our testing data after training our ML model
2) We were able to problem solve through most roadblocks that we encountered using AWS as beginners

What we learned

1) How to use AWS services in conjunction with eachother (S3, SageMaker)
2) How to utilize APIs to connect the frontend to the backend of our app
3) Critical thinking skills to adjust ML model features to improvee accuracy

What's next for Arena Insights

1) Continue tweaking the model to account for different factors to improve testing accuracy
2) Use the predicted attendance values to make recommendations for arena logisitics

Share this project:

Updates