Using Machine Learning to Predict Playground Usage

Student Capstone Project

Children are spending more time in front of screens and less time playing outside – the new normal for a digital generation. But this has brought parents face to face with a challenging question: “how can we get kids more active outdoors?”

As an answer to this question, Biba - a smart playground company based in Vancouver, British Columbia - has developed a collection of mobile games for families to use at local playgrounds in an effort to get kids more physically active. But while park and playground owners are looking to benefit the communities they serve with new opportunities for physical activity, they are also looking to understand more about how their playgrounds are actually being used. After all, additional data around playground usage allows park managers to make meaningful decisions regarding the management of existing playgrounds and even the planning of new ones. As a baseline, this means park owners need to know how many families use playgrounds on a monthly basis - a difficult metric to access.

In order to get a sense of how many families use playgrounds across North America, a team of UBC Master of Data Science Vancouver students worked with Biba to build several machine learning models to estimate the number of monthly sessions at particular playgrounds within a given month. These models were based on data collected across 2506 playgrounds in the United States from two different sources: time-invariant data (i.e. location data, demographic data) and time-variant data (i.e. data collected from Biba’s mobile apps during gameplay and weather data).

The various models they worked on provided a solid foundation for Biba’s in-house data science team to build on. They handed over a populated GitHub repository with a data analysis pipeline that made it easy for the Biba team to work with. These results allowed further critical data processing decisions to be made that lead to enhanced predictive power for a data product that is anticipating release in 2020/2021.

Explore MDS Vancouver Explore Other Data in Action Stories