Continue to EdScoop.com

Boston school bus project underscores the value of data analytics

Designing cost-saving bus routes was possible through initiative with SAS, say officials from Boston Public Schools.

Richard W. Walker
Bio
Richard W. Walker

Richard W. Walker is a freelance writer based in Maryland who has been covering issues and trends in government and public sector technology for mo...

(Getty Images)

When leaders of the Boston Public Schools (BPS) decided last year to revamp their large, complex and inefficient bus system, they turned to advanced analytics and a crowd-sourcing challenge to tackle the longstanding problem.

The school system collaborated on the data-driven initiative with SAS, a North Carolina-based provider of business intelligence and data management software and services, as well as DataKind, a nonprofit devoted to helping connect institutions with pro-bono support. BPS also staged a “hackathon” transportation challenge to help solve the problem. The focus was to improve the efficiency of the BPS bus system, reduce the number of stops, cut transportation costs and reduce bus traffic on busy city streets.

BPS is the oldest public school system in the United States, and its transportation system is one of the largest. Last year, it carried 25,000 students via 650 buses across 45,000 miles. Within this system, BPS buses made more than 20,000 unique stops at nearly 5,000 different locations every day. A BPS official said the main problem was empty buses.

According to Jinxin Yi, a senior manager for advanced analytics in the SAS Operations Research Center of Excellence, the company's role in the project was “a perfect nexus.”

“We were brought in not only due to our analytical depth, but because of our commitment to education and using data for good,” he told EdScoop. “SAS worked with BPS to identify critical steps to improve their current bus routing system, determine the data needed to build out optimization models for each step and report on insights and potential areas for improvement.”

Beginning in August 2016, Yi said, SAS analysts collected a huge amount of data, including student information such as grades, neighborhoods, longitude/latitude coordinates of their home addresses, current stop assignments and the schools they are attending; bus stop information such as longitude/latitude coordinates; school information such as location and address, longitude/latitude coordinates and opening-bell times; and bus information such as the types and capacities of buses.

SAS analysts then used the data to build several models to optimize student bus-stop assignments and bus schedules for all bus routes, he said.

“We analyzed the student/stop assignments based on various settings, such as maximum walking distance by grades, maximum walking distance by neighborhoods and maximum number of students per stop. BPS used these results to determine which settings would work best for them,” Yi said.

DataKind, which focuses on "data-driven projects for social impact," assisted in sourcing and setting up the project for SAS, he said.

“They first worked with Boston Public Schools to better understand their problems and needs, and determine how and where data science could best be applied to address these challenges,” he said.

SAS analytics’ research showed that the district could operate more efficiently and cost-effectively by reducing the number of bus stops anywhere from 20 percent to 50 percent, according to SAS. The analysis also allowed BPS to think differently about stop assignments.

Previously, students were assigned to a stop within a mile of their homes. Applying analytics, SAS was able to create more walk-to-stop maximums for each student by factoring in variables such as the safety of that student’s neighborhood. This approach, combined with a reduction in the number of bus stops, allowed BPS to deliver a simpler, more effective transportation system for its students, SAS said.

SAS’s analytics inspired BPS to organize a hackathon-style event called the Transportation Challenge, which ran from April 1 to mid-July this year. Data scientists, optimization experts and mathematicians from such places as Google, Microsoft, Uber, Harvard, Georgia Tech and MIT participated in the challenge. Its purpose was “to generate efficient solutions that could help to reduce our transportation budget,” said John Hanlon, chief of operations at BPS.

“One of the primary goals of the Transportation Challenge was to leverage some of the smartest optimization experts in the world to help us build efficient routes for our buses, which could then save money that could then be reinvested in the classrooms,” Hanlon told EdScoop.

A team from MIT won the challenge. Their work helped BPS eliminate 50 bus routes this year, said Hanlon, who added that the school system expects to realize $3 million to $5 million in savings from the effort.

“The fact that we’re taking 50 buses off the road every day means that we’re reducing our carbon footprint by 20,000 pounds of carbon emissions every day,” he said. “We’re also reducing the number of miles driven by our buses in Boston by a million miles every year.”

Despite adjustments to routing, the average walk-to-stop per student is the same as last year, as is the average commute time, Hanlon said.

-In this Story-

Education IT News, Computer Science, Boston Public Schools, SAS, DataKind, MIT, Transportation Challenge, Harvard

Join the Conversation