MIT's Optimization Algorithm Cuts Boston Public Schools' Bus Costs by $5 Million

In 2016, Boston Public Schools launched a public "Transportation Challenge" inviting outside teams to redesign its school bus routes; a team from MIT's Operations Research Center, led by Professor Dimitris Bertsimas, won the competition.

Redesign bus routing to cut costs, mileage and emissions while maintaining service, using an optimization algorithm that combines traffic, enrollment and address data.

Boston Public Schools piloted the winning algorithm for the 2017-18 school year, assigning students to stops, combining stops into routes and allocating vehicles using Google Maps traffic data; the methodology was later peer-reviewed and published in the journal Interfaces (2020).

The district eliminated about 50 routes (roughly 8% of the fleet), cut about 1 million miles of driving, reduced daily CO2 emissions by an estimated 20,000 pounds, and saved between $3 million and $5 million annually, while cutting route-planning time from thousands of staff-hours to about 30 minutes.

The tool is an operations-research optimization system rather than a machine-learning model, and its savings are specific to Boston's route network and enrollment patterns; other districts would need their own re-optimization, and issues such as driver shortages remain separate from the routing math.