Case Study: TransLink improves bus departure predictions by 74% with Microsoft Power BI and Azure Machine Learning

Metro Vancouver transportation authority improves bus departure estimates by 74 percent with Azure Machine Learning

TransLink, the Metro Vancouver transportation authority, faced growing rider frustration as increased traffic and ridership made scheduled bus departure times unreliable. To give customers accurate, real-time departure estimates the agency leveraged Microsoft technologies—most notably Azure Machine Learning for predictive modeling and Microsoft Power BI for reporting—working with partner T4G to modernize its prediction system.

Using Azure Machine Learning and MLOps, TransLink trained more than 18,000 stop- and segment-level models, automated training pipelines with Azure DevOps and Azure Data Factory, and used Microsoft Power BI to monitor model performance. The result: departure predictions improved by 74 percent, riders’ wait time fell by 50 percent, and the share of riders waiting more than five minutes dropped from 18% to 4%, delivering a markedly more reliable transit experience.

Translink

Maria Su

Director of Research and Analytics