Chicago Transportation Network Providers Trips (Uber/Lyft/Via)

Chicago Transportation Network Providers Trips (Uber/Lyft/Via) is Chicago's trip-level record of its ride-hail market: every ride dispatched by the companies the city regulates under its Transportation Network Provider rules - the ordinance-speak for Uber, Lyft, Via and their peers - reported trip by trip because the city requires it. Coverage opens in November 2018 and runs continuously across three archives: November 2018 through December 2022, calendar 2023-2024, and January 2025 onward, where the current archive alone holds roughly 139 million rows.

Each row is one completed ride. When it started and ended (rounded to fifteen minutes), how long it took, how far it went, where the passenger was picked up and dropped off down to census tract and community area, and the complete money picture: fare, tip, taxes and fees, and the total. Sharing detail travels with every trip - whether the rider agreed to pool, whether a match actually happened, and how many rides shared the vehicle.

That makes this the sharpest single view of app-based mobility in any major American Midwest market: an entire city's ride-hail activity observed at one-row-per-trip resolution, year after year, with no sampling layer between you and the rides.

Get a sample of this dataset and judge the rows before anything ships. It sits inside our passenger ground transportation data hub alongside the rest of the industry's cataloged sources.

Illustrative rows in the delivered shape, grouped by role:

trip_id                             trip_start_timestamp  trip_end_timestamp
001eb95052a52c28e7ec84bcbdfdea7b25a75a0e  2026-06-30T23:45:00.000  2026-06-30T23:45:00.000

trip_seconds  trip_miles  percent_time_chicago  percent_distance_chicago
490           2.65        0.998                 1

pickup_census_tract  dropoff_census_tract  pickup_community_area  dropoff_community_area
17031833000          17031241400           28                     24

pickup_centroid_latitude  pickup_centroid_longitude  dropoff_centroid_latitude  dropoff_centroid_longitude
41.8852813201             -87.6572331997             41.906025969               -87.66351755

fare   tip  additional_charges  trip_total
12.5   5    5.19                22.69

shared_trip_authorized  shared_trip_match  trips_pooled
false                   false              1

Read the anatomy rather than the digits - every value above is the documented example for its column, not a live observation. Three things stand out once rows are laid out this way. First, geography comes at two resolutions side by side: census tracts where privacy allows, community areas and centroid points everywhere else, so no trip ever arrives locationless. Second, the money splits into fare, tip and additional_charges before summing to trip_total, which lets platform revenue, driver-side tipping and statutory taxes and fees be separated without guesswork. Third, the sharing trio tells the whole pooled-ride story: intent (shared_trip_authorized), outcome (shared_trip_match) and load (trips_pooled).

All twenty-four cataloged fields are verified - every name, type, definition and example below was checked against the official per-archive documentation rather than inferred from column headers, which is why this dataset scores 9 on our field-documentation rubric against a catalog average of 7.81 across 1,744 datasets.

Four column groups do most analytical work. The timestamp pair fixes each ride to its quarter hour, which is exactly the resolution late-night surge, event egress and airport-run analyses need anyway. The geography quartet pairs census tracts with community areas, so a Wicker Park-to-Loop flow matrix and a 77-neighborhood demand panel come off the same rows. The money quartet separates fare from tip from additional_charges before trip_total closes the arithmetic - useful precisely because the parts behave differently: fares snap to $2.50 steps, tips to $1.00 steps with cash never recorded, while the charges column carries the taxes and fees. And the sharing trio distinguishes riders who opted into pooling from rides that actually matched.

The 2025-onward archive adds a location-quality upgrade worth knowing about: percent_time_chicago and percent_distance_chicago state directly how much of each ride happened inside the city, turning cross-border suburban trips into filterable segments instead of judgment calls.

Additional fields on request. Column availability shifts by era - the percent-Chicago measures and shared_trip_match exist only from January 2025 forward - and era-by-era column manifests are documented against delivered extracts rather than promised blind. Ask for the sample and the complete per-archive column list arrives with it.

• Geographic: the City of Chicago plus the surrounding region, with endpoints coded to census tracts and to Chicago's 77 community areas. Where tract identity is withheld for privacy, the tract's centroid point stands in - so every trip keeps a usable location even when precision is deliberately blurred.
• Temporal: November 2018 straight through to the present, held as three archives (2018-2022, 2023-2024, 2025-onward) that join cleanly on identical field names, observed through June 2026. Eight years of history means every month-of-year has multiple precedents for seasonality baselines.
• Granularity: one row per individual ride-hail trip - no sampling, no pre-aggregation. Scale runs to roughly 139 million rows in the 2025-onward archive alone, with multi-hundred-million-row earlier tables and a multi-hundred-gigabyte footprint overall.

The ladder matters. The same base serves a single-community-area pickup audit and a citywide platform-share panel without reconciliation, because both read off the same trip-level rows.

API, files, or your warehouse. Daily, weekly, or hourly.

You choose the slice - one community area's Friday-night history, one corridor's fare curve since 2019, pooled-ride uptake across the whole city - and the shape: flat files for analysts, a feed for running pipelines, or landed tables in your own warehouse. Most teams take the eight-year history once as a backfill and keep fresh trips rotating in on whatever cadence their models expect. Datadory normalizes the three-era quirks described above (the 2025-only columns, the tract-versus-centroid geography, the binned fare and tip steps) so the extract arrives analysis-ready in one consistent frame. Get a sample of this dataset scoped to your neighborhoods and months.

• Cab-versus-app market share - ride-hail volumes here against the parallel Chicago taxi series turn "Uber ate the cab business in Chicago" from anecdote into a measurable monthly series; see competitive intel product teams use cases.
• Demand, ETA and fare modeling - trip-level durations, distances and quarter-hour timestamps across eight years are a ready training substrate for forecasting models that need real city physics; see data scientists use cases.
• Neighborhood and corridor economics - pickup and dropoff counts by community area act as a high-frequency proxy for commercial activity, nightlife draw and office reoccupancy, feeding mobility and real-estate theses; see investors quants use cases.
• Shared-mobility uptake studies - the authorized-versus-matched-versus-pooled trio is one of the few places pooling behavior can be measured at the ride level rather than inferred from press releases.
• Policy evaluation - the taxes-and-fees column carries the per-trip levies separately, giving before/after designs a clean outcome variable for downtown congestion measures introduced in 2020 and anything since.
• Citable journalism and research - every claim traces back to an individually observed, regulator-reported trip.

Data scientists and ML engineers (relevance 3 of 3) get hundreds of millions of labeled trip observations with built-in seasonality and a geographic hierarchy ready-made for spatial features. Market researchers and consultants (3 of 3) quantify ride-hail penetration, fare levels and neighborhood demand from regulator-reported rows rather than vendor surveys. Journalists, academics and students (3 of 3) cite a mandate-backed trip record whose provenance is a municipal reporting requirement - the strongest kind of citation. Competitive intelligence teams (2 of 3) track app-based growth against the city's taxi series as evidence of modal shift. Urban planners and policy analysts (2 of 3) read curb pressure, airport flows and congestion-fee effects directly off endpoint geography. Investors and quant researchers (2 of 3) back mobility, logistics and real-estate theses on point-in-time trip files spanning eight years.

Within US ground transport data, this dataset owns ride-hail isolation: it is the major-market series that separates app-based trips from street-hail cabs by construction, because only TNP-reported rides appear here. Chicago Taxi Trips is its deliberate mirror - same city, same era, cab side - and the head-to-head is laid out on our Chicago taxi versus TNP comparison. Pairing them yields a complete two-mode panel for one city.

NYC TLC Trip Record Data is the structural sibling one metro over: trip-level high-volume for-hire records including Uber and Lyft, deeper history and richer pay fields. Together they make a two-city ride-hail panel; the usual stack adds the taxi feeds for the cab layer and national survey sources for context. Where everything ranks sits on our best passenger-ground-transportation datasets list.

Notes worth having in hand before you rely on it:

• Three archives, one series - coverage is continuous but physically split at the end of 2022 and end of 2024; join on field names and the series behaves as one, but always state which archives contributed to any total.
• Quarter-hour timestamps - start and end times round to fifteen minutes; second-level dispatch analysis needs the duration field instead of subtracting timestamps.
• Money is binned, not exact - fares round to $2.50, tips to $1.00 and cash tips never appear; treat monetary fields as banded distributions and say so in any per-trip revenue claim.
• Tracts yield to centroids - some trips arrive with the tract withheld and a centroid point standing in; point-level joins should expect the centroid fallback rather than treating it as noise.
• Newer columns cut both ways - the percent-Chicago shares and shared_trip_match exist only from January 2025 forward, so panels spanning earlier eras need those columns treated as missing, not zero.
• Count the big archives carefully - the 2025-onward row count is well established at roughly 139 million, but the two earlier archives were not fully enumerated during verification; quote combined totals as estimates until they are.Pages that pair well with this one: the passenger ground transportation data hub for the pooled industry view, the Chicago taxi versus TNP comparison for the two-mode framing, and the persona pages for how data science and developer teams put these rows to work.