Passenger Ground Transportation Data Provider · Head-to-head
Chicago Taxi Trips vs Chicago Transportation Network Providers Trips (Uber/Lyft/Via)
Which passenger ground transportation data provider data fits your job: Chicago Taxi Trips, or Chicago Transportation Network Providers Trips. API, files, or your warehouse. Daily, weekly, or hourly.
Chicago Taxi Trips
Chicago Transportation Network Providers Trips (Uber/Lyft/Via)
Where the fields line up
17 shared fields — join on these.
| Field | Chicago Taxi Trips | Chicago Transportation Network Providers Trips |
|---|---|---|
trip_id | documented | documented |
trip_start_timestamp | documented | documented |
trip_end_timestamp | documented | documented |
trip_seconds | documented | documented |
trip_miles | documented | documented |
pickup_census_tract | documented | documented |
dropoff_census_tract | documented | documented |
pickup_community_area | documented | documented |
dropoff_community_area | documented | documented |
fare | documented | documented |
trip_total | documented | documented |
pickup_centroid_latitude | documented | documented |
pickup_centroid_longitude | documented | documented |
pickup_centroid_location | documented | documented |
dropoff_centroid_latitude | documented | documented |
dropoff_centroid_longitude | documented | documented |
dropoff_centroid_location | documented | documented |
Coverage, side by side
| Chicago Taxi Trips | Chicago Transportation Network Providers Trips | |
|---|---|---|
| Geographic | City of Chicago plus surrounding region; census tracts and community areas inside the city, blanks beyond | City of Chicago plus surrounding region; census tracts and community areas inside the city |
| Temporal | 2013-2023 in the legacy archive; January 2024 to present in the successor, observed through August 2026 | November 2018-2022, 2023-2024, and January 2025 to present, observed through June 2026 |
| Granularity | One row per individual taxi trip | One row per individual ride-hail trip |
What each contains
They tie on 2 attributes. Pick by fit, not by loyalty.
| Chicago Taxi Trips | Chicago Transportation Network Providers Trips | |
|---|---|---|
| Publisher | City of Chicago, via the Chicago Data Portal | City of Chicago, via the Chicago Data Portal |
| Reporting basis | Trips reported to the city's taxi regulator | — |
| Unit of analysis | One row per individual taxi trip | One row per individual ride-hail trip |
| Geographic coverage | City of Chicago plus surrounding region; census tracts and community areas inside the city, blanks beyond | City of Chicago plus surrounding region; census tracts and community areas inside the city |
| Temporal reach | 2013-2023 in the legacy archive; January 2024 to present in the successor, observed through August 2026 | November 2018-2022, 2023-2024, and January 2025 to present, observed through June 2026 |
| Scale | About 1.7 billion rows for 2013-2023 plus roughly 17.5 million since 2024; multi-hundred-gigabyte fully materialized | Roughly 139 million rows in the 2025+ archive alone atop multi-hundred-million-row earlier tables; multi-hundred-gigabyte overall |
| Documented fields | 23 | 24 |
| Datadory rubric | 9/10, field definitions verified | 9/10, field definitions verified |
| Best for | Fare precision, tipping behavior, payment mix, company concentration, pre-2018 history | Current mode share, pooled-ride economics, taxes and fees, within-Chicago trip shares |
What each does better
Chicago Taxi Trips
Money at full resolution. fare, tips, tolls, extras and trip_total arrive unrounded - a $9.50 fare with $2.90 tip totals $12.90 exactly as paid. The ride-hail side bins its money before publication: fare to the nearest $2.50, tip to the nearest $1.00. Anyone modeling fare elasticity, tip rates or per-mile yield gets real numbers here and stepped approximations there.
Payment context and company identity. payment_type separates card from cash from mobile - essential because cash tips generally never get recorded - and company names the operator outright (the sample row runs on Flash Cab). Fleet-level work like consolidation tracking or utilization studies has no ride-hail equivalent, because platforms are never named per trip.
A stable pseudonymous fleet key. taxi_id is a consistent hash per medallion, so vehicle-level histories can be built without ever publishing a medallion number. That is the quiet achievement of this dictionary - longitudinal fleet analytics under a privacy constraint.
Deeper history. Trips run from 2013 forward, giving thirteen-plus years of continuous coverage. The ride-hail series begins November 2018, when ordinance-mandated reporting started. Any seasonality claim that wants a decade of precedents only finds them here.
Chicago Transportation Network Providers Trips
The sharing story, told in three fields. shared_trip_authorized records whether the customer agreed to pool, shared_trip_match whether a match actually happened, and trips_pooled how many rides shared the vehicle's cycle. Pooling economics - uptake by neighborhood, discount versus occupancy, the real supply curve of shared rides - simply have no taxi-side counterpart.
Platform-era scale. Roughly 139 million rows in the 2025-onward archive alone, stacked on multi-hundred-million-row 2018-2022 and 2023-2024 tables. Where the taxi corpus thins after 2016, ride-hail thickens - if you want today's Chicago demand, most of it rides in this table.
A within-Chicago ruler. percent_time_chicago and percent_distance_chicago grade every trip by how much of it happened inside the city - fields the taxi dictionary does not carry, and the difference between counting regional activity and counting city activity.
Cleaner charge separation. Money splits into fare, tip and additional_charges, isolating statutory taxes and fees from the meter and the driver's tip. Even binned, that three-way split supports tax-burden and fee-incidence analysis the taxi columns cannot isolate as cleanly.
Where they're equivalent
One grain, stated identically. One row per individual trip on both sides - no sampling, no pre-aggregation - which is why a single query pattern serves a late-night community-area audit and a citywide monthly panel against either table.
The same geography kit. Pickup and dropoff census tracts, pickup and dropoff community areas, and centroid latitude-longitude pairs appear in both dictionaries. Privacy behaves the same way too: tracts suppressed on some trips, with the tract or community-area centroid standing in so no trip arrives locationless.
Identical time handling. Start and end timestamps rounded to the nearest fifteen minutes on both sides, with trip_seconds preserving true duration. Minute-level dwell math is off the table everywhere; hour-of-day and day-of-week curves survive intact everywhere.
Same grade, same verification. Both score 9/10 with verified field definitions from the same research pass, and both report daily refresh cadence upstream.
The verdict
Verdict: sample both, pick by fit - the question decides, not the leaderboard.
Take the taxi series when precision matters more than volume: unrounded fares and tips for pricing or tip-rate models, payment-type context, tolls and extras itemized, named operators for market-concentration work, or history before 2018.
Take the TNP series when the question lives in the platform era: mode-share and demand modeling at current volumes, pooled-ride uptake from the authorization-match-count trio, taxes-and-fees incidence from the charge split, or the share of each trip inside Chicago.
Three quick tests settle most cases. Need exact dollars per trip? Only taxis publish them. Need shared-ride behavior? Only ride-hail carries the flags. Modeling how Chicagoans moved across the last decade, cab era into app era? That is the both-of-them case, and it is the common one.
Sample both, pick by fit. See Chicago Taxi Trips · See Chicago Transportation Network Providers Trips
Or take both in one feed
They stack along one timeline, and the join is a column mapping rather than a miracle. Because both dictionaries carry fifteen-minute timestamps, duration, distance, census tracts, community areas and centroid points, a single pickup-tract by month panel covering both modes assembles directly - the canonical taxi-versus-ride-hail modal competition study, run inside one major US market without stitching sources.
Two practical notes from the records. Mind the money: taxi fares are exact while ride-hail fares move in $2.50 steps and tips in $1.00 steps, so aggregate fare comparisons should sit at means and medians, not per-trip deltas. And mind the blind spots: cash tips go unrecorded on both sides, and tips versus tip is the same quantity under two names - harmonize before summing.
Or take both in one feed. Datadory normalizes each record to its documented dictionary - 23 fields on the taxi side, 24 on the ride-hail side - aligns the overlapping columns, attaches sample rows for validation, and ships them beside the rest of the passenger ground transportation catalog - delivered daily, weekly, or hourly, your call.
API, files, or your warehouse. Daily, weekly, or hourly.
Fair questions
Is the Chicago taxi dataset better than the Uber/Lyft/Via dataset?
Different instruments, tied on craft - both score 9/10. The taxi series wins when you need unrounded fares, payment type, itemized tolls and extras, or a named operating company; the TNP series wins when you need platform-era scale, pooled-ride flags and the share of each trip spent inside Chicago.
Can the Chicago taxi and TNP datasets be joined?
Yes - it is their best trick. Both carry trip identifiers, fifteen-minute timestamps, duration, distance, census tracts, community areas, centroid coordinates and a fare breakdown, so a pickup-tract-by-month panel covering both modes assembles without any crosswalk table.
How far back does each Chicago trip dataset reach?
Taxis reach further: 2013 through December 2023 in the historical archive plus January 2024 onward in its successor. Ride-hail starts November 2018, when Chicago first required Transportation Network Providers to report every trip, held across three archives running to the present.
Which Chicago trip dataset is bigger?
Ride-hail, decisively. The 2025-onward TNP archive alone holds roughly 139 million rows atop multi-hundred-million-row earlier tables, while the taxi side counts about 1.7 billion rows over thirteen years but only around 17.5 million since 2024 - the mode-share shift made literal in row counts.
How current can Chicago taxi and ride-hail data be?
As current as your project needs: Datadory delivers either dataset daily, weekly, or hourly - your call. Content-wise both run to the present, with taxi trips observed into August 2026 and ride-hail trips observed through June 2026.
Can Datadory deliver both datasets together?
Yes. Sample both and pick by fit, or take both in one feed - normalized to their documented dictionaries (23 fields on the taxi side, 24 on the ride-hail side), harmonized on the overlapping columns, and validated against sample rows before anything ships.