Garage interns bring machine learning to TransLink

, ,

Imagine sitting down with the head of Metro Vancouver’s transportation network to talk about the role machine learning could play in the organization. Cool, right? Now imagine you’re an intern.

In May of 2017, two dozen university students from across Canada began a four-month entrepreneurial internship at Microsoft Vancouver. It was the second of two cohorts that the global development centre recruits each year, and the first group to be pitched a combination of corporate and community projects. Enter TransLink.

“TransLink came to us with a few years of ridership data that they were willing to share,” explained Stéphane Morichère-Matte, Talent Builder for Vancouver’s Microsoft Garage Internship Program. “It was an opportunity to have the students explore machine learning and show the organization how it could impact their bottom line.”

With a team of six, the interns began working with TransLink and a group of Microsoft mentors. They utilized Xamarin, Azure App Services, including Azure ML, Visual Studio Team Services and more, to build a working prototype for the iOS platform to determine options for applications that customers might find useful.

“This was a great opportunity for us to work with bright, young talent,” noted Kurt Pregler, Chief Information Officer for TransLink. “By having the opportunity to work with these insightful and energetic interns, we were able to gain insights that would have taken longer to learn internally.”

The project focused on creating a platform that could determine bus ‘crowdedness’ using machine learning. With the support of Microsoft’s Machine Learning team in Redmond, the interns injected TransLink data in AzureML and ran many simulations to build predictive models of bus crowdedness. Armed with this predictive model, the interns built a mobile app to allow people to plan their trips based on how crowded a bus might be at different times of day, hoping to change ridership habits.

The work done by the interns was turned over to TransLink at the end of the project, giving the organization an avenue for future customer facing applications. For Microsoft, it was an opportunity to develop a closer relationship with team, increase awareness of Azure services and work directly with TransLink on managing growth through machine learning and application lifecycle management.

“While the team didn’t ship a finished app, the work they did is being used internally by TransLink and will inform future projects,” said Morichère-Matte. “It was a valuable community builder that challenged the interns and provided them with solid experience in machine learning.”

 

SaveSave