Moving the needle on ‘moving people’
Here's how RTD approaches the data that will determine the future of public transportation in Denver.
After spending much of the past 15 years building as much as it could as fast as it could, RTD is as focused as ever on its core mission: moving people.
The 2004 approval of the FasTracks ballot measure set in motion a decade and a half of construction and expansion of light-rail and commuter-rail lines in Denver. Opening a new line and extending another in 2019 added to RTD’s “50 years of moving people” celebration as the organization marks a half-century of transporting Coloradans.
Denver’s public transportation system had the 11th-highest ridership of any metropolitan area in the country last year, according to data from the American Public Transportation Association, but the number of annual boardings dropped under 100 million last year for the first time in several years. Denver’s drop reflects a sharp national decrease in public transit usage, about 8% in total ridership from 2014-18.
“We’ve had a great build-out, and that’s where our efforts have been concentrated, these huge infrastructure projects. Now we need to fill those resources up. Now that the lines are built and have opened, we need to focus on ridership,” said Dennis Yaklich, manager of data science and analytics at RTD.
Yaklich leads a team created last year to help get the most out of RTD’s offerings. His five-person data science and analytics crew includes experts in database administration, business intelligence, economics, accounting and project management.
Together, they’re helping RTD work toward an internationally recognized asset- management certification, which would help the organization meet federal compliance requirements and demonstrate a commitment to efficient use of taxpayer dollars.
RTD is also working with a consultant on a comprehensive operations assessment to help figure out what constituents want next, such as more bus rapid-transit lanes. That assessment will include a significant community-engagement component that helps RTD develop a road map for where to go next based on the needs of the region.
Yaklich brings a dynamic perspective to data science: He has a psychology degree and management minor from Metropolitan State University of Denver, where he teaches Analytics and Statistics for Marketing Decisions in the Master of Business Administration program.
He first joined RTD nine years ago to do market research, carefully designing surveys for representative population samples of up to 10,000 participants. Strategically collecting feedback from constituents can be a difficult task, as human brains are hard-wired to remember negative experiences more than positive ones.
“We’re built to record threats or negative events because the consequences are so much higher. In order for the species to make it where it is, we needed to be better at recognizing threats,” Yaklich said. “For instance, safety is a funny thing to measure. People can tell you a story of when they felt unsafe, but all those times you felt safe disappear from your memory.”
A lot of the work his team does informs executive decisions about what to prioritize – riders always say safety is important, as is on-time performance, but things such as the exterior appearance of a vehicle or various bells and whistles take a back seat when a taxpayer-funded organization decides what to do next.
“You can’t be Chick-fil-A and a steakhouse. You have to focus on your core competencies as an organization to be effective and efficient. That means you can’t do everything,” Yaklich said. “You’ll hear people say, ‘What you need is Wi-Fi on the buses.’ No, we need to show up on time and take people where they want to be.”
Optimizing bus routes and schedules is an ongoing project for RTD, and Yaklich wants to bring a new perspective to the process: machine learning. He compares it to programming a computer to find the fastest path to complete “Super Mario Bros.” – simulating the game thousands of times to find the optimal routes and jumps. If an algorithm determines buses would arrive on time more often if they came every 12 minutes as opposed to 15, or that they should go different routes on different days of the week because of predictable traffic, then why not let data help decide instead of just guessing what works best?
“People want to be able to predict human behavior, and they’re not good at it. It’s not intuitive. There’s a concept of expert intuition, and they’re wrong, almost always. If you can help people make better decisions with data, you can make the world better,” Yaklich said.