Projects

Got a Ride? Ridesharing Equity in Chicago

Fall 2021
Introduction to Urban Data Informatics
Columbia GSAPP
Prof. Boyeong Hong

I expected Uber and Lyft pickups to cluster in white, wealthy Chicago neighborhoods. 42 million trips later, the data said no - the pickups go where the trips are, which mostly means downtown.

Scroll

The Question

Hypothesis
Rideshare access tracks
income and race - rejected

42M trips · 14 months
77 Community Areas
EPSG:4326

Uber and Lyft are now a meaningful share of Chicago's transit mix. Whether they reach the neighborhoods that need them is a separate question.

Guess: pickups skew toward wealthier, whiter community areas. The test set: 42 million anonymized rideshare trips from the City of Chicago's TNP dataset (every licensed rideshare pickup, Oct 2020 – Nov 2021), joined against Census income and race data across all 77 Community Areas.

Follow the Pickups

Five stops across 42 million trips. Scroll.

The whole city

Every rideshare pickup Chicago logged from October 2020 to November 2021, shaded by volume: 41.9 million pickups across 77 community areas. Most of the map is pale. The dark core isn't.

Darker teal = more pickups (log scale, 25K to 6.0M)

Six areas take nearly half

The Near North Side, Near West Side, Lake View, the Loop, West Town, and Lincoln Park account for 45.6% of every pickup in the city. The Near North Side alone logs 6.0 million - jobs, nightlife, hotels, and two train stations' worth of demand.

Outlined: the six busiest community areas

The South and West Sides ride too

If pickups skewed white and wealthy, the map would go quiet here. It doesn't. Austin, 94% nonwhite, logs 938K pickups. South Shore logs 605K, Englewood 317K, Washington Park 187K. Demand follows jobs and transit gaps, not demographics.

Now shaded by % nonwhite residents (purple = higher); outlined areas named above

Forest Glen barely calls a ride

The city's second-richest community area: $124K median income, 66K pickups in 14 months. The Near North Side, at a similar income, logged 90 times more. Up here almost everyone owns a car - wealth predicts car ownership, and car ownership replaces rideshare.

Back to pickup shading; Forest Glen outlined

Pickups follow trips, not people

Nine community areas top $100K median income, and their pickup counts run from 41K to 6.0 million. Income doesn't predict demand; density and destinations do. The hypothesis - that rideshare quietly favors richer, whiter neighborhoods - doesn't survive the map.

The scatter below makes the same case in one frame

Pickups vs. Income

If pickups tracked income, this cloud would climb to the right. It doesn't.
Median household income against rideshare pickups (Oct 2020 to Nov 2021, log scale), colored by percent nonwhite population. Downtown carries the cloud - Near North Side alone logs 6.0M pickups - while Forest Glen, one of the city's richest areas, barely registers 66K. Car ownership, not income, drives the gap. Hover any dot.

Key Numbers

42M
Trips Analyzed
77
Community Areas
14 mo
Time Window

The Same Map, Twice

Why a single-variable equity test stalls in Chicago
Median income beside white share, both darker-is-higher across the same 77 community areas. The North Side reads dark on both maps, the South and West sides light on both - income and race are so entangled here that any one-variable test of rideshare equity measures both at once.

Findings

  • Hypothesis: disproven. Pickup volume doesn't track income or share of white residents at the Community Area level. Whatever rideshare is doing in Chicago, it isn't quietly skewing toward whiter, richer neighborhoods.
  • High income can mean low demand. One of Chicago's highest-income areas had among the fewest pickups - those residents own cars and drive themselves.
  • Pickups follow trips, not people. The Loop, the Near North Side, and the big employment corridors dominate, full stop. That's where people actually need rides - regardless of who lives there.
  • Income and race are nearly the same map. The two choropleths are visual near-mirrors of each other in Chicago. Any single-variable equity test runs into that wall before it can say much.
This project was completed for Introduction to Urban Data Informatics at Columbia GSAPP (Fall 2021, Professor Boyeong Hong). Data sourced from the City of Chicago TNP dataset and US Census. Notebook and the original interactive Folium bubble maps live on GitHub; an earlier version of this writeup is on Medium.