Objects

A topic you can ask for:

• demographics: Demographics (population, age, households, education)
• income: Income & poverty
• housing: Housing (homes, vacancy, owners and renters, home values, rents)
• labor: Labor market (jobs located in the area, hiring, unemployment and residents' work)
• businesses: Businesses & amenities (places such as grocery stores, bank branches, workplaces, public facilities)
• health: Health (health outcomes, insurance, Medicare, care facilities, shortage areas, food access)
• environment: Environment & hazards (natural hazards, flood zones, regulated sites, drinking water, oil and gas wells)
• incentives: Tax incentives (Opportunity Zones and other areas where tax incentives apply)
• transportation: Transportation (traffic, transit, EV charging, commuting, fatal crashes)
• spending: Consumer spending (what households spend a year, estimated from their incomes)
• timeuse: Time use (how residents spend their week, and how many people are present at each hour)
• education: Education (public schools and colleges)
• broadband: Broadband (home and business internet service offered, its speeds and providers)

The topic's name in plain words.

What the topic covers.

The groups in this topic. You pay one credit per group per place.

False until this topic's data is in a published release.

A location as [longitude, latitude], in that order, in standard GPS coordinates (WGS84). Longitude comes first, so Boston is [-71.06, 42.36].

A location as [longitude, latitude], in that order, in standard GPS coordinates (WGS84). Longitude comes first, so Boston is [-71.06, 42.36].

How far out to go. Up to 50 miles (80 km).

mi (miles), km (kilometers) or m (meters). Default "mi".

A location as [longitude, latitude], in that order, in standard GPS coordinates (WGS84). Longitude comes first, so Boston is [-71.06, 42.36].

How you travel: drive, walk or bike.

Travel time, 1 to 60 minutes. Walking and biking go up to 30.

Polygon (one shape) or MultiPolygon (several separate pieces).

The corner points of the outline, as [longitude, latitude] pairs.

The area's official ID number.

Your own name or ID for the place. We send it back with the results so you can match them up.

The place, in any one of these forms.

Which data release to use (a frozen snapshot, such as 2026.11 or 2026.11.1). Name one to get the same answer every time; latest uses the newest. Default "latest".

The official areas used as building blocks for your shape: block_group (smaller and more detailed, the default) or tract (larger, with smaller margins of error). Default "block_group".

How we split official areas that your shape only partly covers:

• dasymetric (default): by where people actually live, using 2020 Census counts. The most accurate.
• areal: by the share of land your shape covers. Simpler; mostly for comparison.
• centroid: an area counts fully if its center point is inside your shape, and not at all if it isn't. Default "dasymetric".

Also return the same numbers for the county, state or whole country, for comparison, plus where your place ranks. The county and state are the ones holding most of your place's people. Rates and medians are ranked against the census tracts in that county, state or country; counts, such as population, aren't ranked, since they grow with the size of your shape. Free.

Include margins of error (how far the true number could reasonably be from our estimate). On by default. Default true.

Send back the outline of each place, for example to draw it on a map. Default false.

Try to fix small problems in your outlines, such as an edge that crosses itself, instead of returning an error. Default false.

Show what the request would cost without running it. Free. Default false.

A ready-made request by name (GET /presets lists them): its groups and comparisons are added to anything else you ask for. Its suggested area isn't applied for you; send that shape as the place.

The places, up to 100. For more, use /jobs.

Whole topics, such as demographics. You pay one credit for each group in the topic, per place.

Groups of related measures, such as housing.costs. One credit per group per place.

Single measures, such as income.household.median. You pay for the group each one belongs to, once per group per place.

Indices (single scores that combine several measures), such as kadrin.economic_momentum. Three credits each per place.

Optional settings. The defaults suit most uses.

Ask for it by this name, as preset.

Its name in plain words.

What it's for.

The groups it includes, which is what you pay for.

The comparisons it adds (free). One of county, state, us.

The area it suggests around each point, as /jobs' around takes it, such as a 10-minute drive.

What each place costs: one credit per group, plus 2 if the suggested area is a drive-time area.

The number. Empty when there isn't enough data; a warning explains why.

Margin of error, meaning how far the true number could reasonably be from the estimate. The true value is very likely (90% of the time) between value minus moe and value plus moe.

For modeled numbers (estimated by a statistical model rather than a survey), the range the true value very likely falls in, as [low, high].

What the number counts or measures, such as people, dollars or percent.

How the number was combined for your shape:

• sum: added up
• ratio: both parts of a rate or average added up first, then divided
• interpolated_median: the middle value, estimated from the full spread of values
• weighted_mean, population_weighted_mean: an average in which areas with more people count more
• point_count: places, building permits, oil and gas wells or drilling permits counted inside your shape
• point_sum: a figure of the places inside your shape, added up, such as the new homes of building permits, bank deposits or weekday transit departures
• point_change: building or drilling permits inside your shape in the last 12 months, minus the 12 months before; or a figure of the schools inside it, minus the same figure some years before
• point_percent_change: the same change as a percent of the earlier figure
• point_ratio: two figures of the schools inside your shape, each added up, then divided
• point_distance: how far your place's point is from the nearest one, in miles
• block_sum: added up from the smallest census areas (jobs; people and homes near oil and gas wells)
• block_ratio: two counts from the smallest census areas, each added up, then divided, such as the share of homes in a floodplain
• block_share: the share of your place's homes inside a kind of area, such as an Opportunity Zone (by land where nobody has a home)
• containing_unit: the value of the official area your point falls in
• block_change: added up from the smallest census areas in two years, then the earlier year taken from the later (job growth); its vintage gives both years, such as 2018-2023
• block_percent_change: the same change as a percent of the earlier year
• binned_rate_sum, binned_rate_mean: regional or national rates applied to the households or people in your shape, as a total or as an average per household or person (a modeled number, not a count)
• hourly_sum: people at each hour of an average day, worked out from who lives or works in your shape
• hourly_point_sum: students at the schools in your shape at each hour of an average weekday
• hourly_total: people at home, at work and at school in your shape, added up hour by hour
• line_sum: a figure of the major roads in your shape, added up, each road by the share of it inside
• line_mean: the average over the major roads in your shape
• line_max: the highest single stretch of major road in your shape
• line_nearest: the figure for the major road nearest your place's point
• line_distance: how far your place's point is from the nearest major road, in feet

For people present by hour: the number at each hour of an average day, starting with midnight to 1 am. The value above is the day's average.

The smallest kind of area this number is published for. If that's bigger than your shape, a warning says so.

Where the number comes from, such as acs5 (the Census Bureau's American Community Survey, 5-year estimates).

The years the number describes, such as 2020-2024; for building and drilling permits the months, such as 2025-10 to 2026-09; for oil and gas wells the date the state's records describe, such as 2026-10-20, or the earliest and latest where your place reaches states with different dates (2025-07-22 to 2026-10-20).

For numbers we have only for some cities or states, such as building permits, oil and gas wells and layoff notices: the share of your place's homes that are in the cities (or states) whose records we have for this measure (0.2 is 20%). Below 1, the number counts only that part and the warning partial_coverage says so; at 0 there's no number.

The same number for the county, state and country, for comparison.

Where your place ranks against the census tracts (neighborhood-sized areas) in the county, state or whole country: 80 means higher than 80% of them. Only for rates and medians.

The score.

The range the score can take.

The version of the index recipe used.

Where your place ranks against other areas.

• high_cv: less reliable, because few households were surveyed here. Treat it as a rough guide.
• coarser_resolution: only published for larger areas, such as counties, so it's worked out from them and is less precise for your shape.
• partial_coverage: data is missing for part of your shape or for some years (for building permits and oil and gas wells, coverage says how much is there).
• suppressed: the source withheld this number, usually to protect people's privacy.
• approximate_moe: the margin of error comes from a standard shortcut formula, so treat it as close rather than exact.
• top_coded: the middle value is above the highest amount the source reports separately, so the true median is this number or more (for example, $250,000 or more).
• bottom_coded: the middle value is below the lowest amount the source reports separately, so the true median is this number or less.
• method_disagreement: this number depends heavily on where people live inside the areas your shape cuts through, so treat it as approximate.
• no_base: there were none at the start of the period (such as no jobs in 2018), so there's no percent change.
• no_roads: there are no major roads in the area, so there's no number.
• estimated_traffic: this road's traffic wasn't counted this year, or its state doesn't say when it was, so it's the state's estimate from an earlier count.

The measure the warning is about.

The warning in plain words.

The name or ID you gave the place.

Facts about the shape itself.

The numbers you asked for, keyed by measure ID, such as income.household.median.

The index scores you asked for, keyed by index ID.

How official areas that your shape only partly covers were split.

Anything to know before relying on these numbers.

If this place failed, what went wrong. Failed places aren't charged.

A short ID for the source, such as acs5.

Its full name, such as U.S. Census Bureau, American Community Survey 5-year estimates.

The years the data describes.

The terms the source publishes its data under.

Where to read more about the source.

An ID for this request. Include it if you contact support.

The data release that answered this request.

Credits this request used.

One result per place, in the order you sent them.

Every source behind the numbers in this response.

Total credits the job would use if every place works.

Rows with a place we can use.

Rows in the file.

Your file's columns, in order.

Which of your columns we read as what, such as {"lat": "Latitude"}.

Your file's first rows, as written.

The first rows we can't use, and why.

How many rows we can't use. They aren't charged.

Total credits the request would use.

The cost for each place.

The measure's ID.

Its name in plain words.

A topic you can ask for:

• demographics: Demographics (population, age, households, education)
• income: Income & poverty
• housing: Housing (homes, vacancy, owners and renters, home values, rents)
• labor: Labor market (jobs located in the area, hiring, unemployment and residents' work)
• businesses: Businesses & amenities (places such as grocery stores, bank branches, workplaces, public facilities)
• health: Health (health outcomes, insurance, Medicare, care facilities, shortage areas, food access)
• environment: Environment & hazards (natural hazards, flood zones, regulated sites, drinking water, oil and gas wells)
• incentives: Tax incentives (Opportunity Zones and other areas where tax incentives apply)
• transportation: Transportation (traffic, transit, EV charging, commuting, fatal crashes)
• spending: Consumer spending (what households spend a year, estimated from their incomes)
• timeuse: Time use (how residents spend their week, and how many people are present at each hour)
• education: Education (public schools and colleges)
• broadband: Broadband (home and business internet service offered, its speeds and providers)

The group it belongs to, which is what you pay for.

What kind of number it is:

• count: how many
• rate: a share or percent
• median: the middle value; half are above it and half below
• average: the total divided by the number of people or households
• modeled: estimated by a statistical model rather than a survey
• point_count: places, permits or wells counted, such as grocery stores, solar panels installed or active oil and gas wells
• change: how a count moved over a number of years, such as jobs gained over the last 5 years
• distance: how far away the nearest one is, such as the nearest Superfund site
• road: a figure of the major roads in the area, such as the traffic on the busiest one

What it's measured in, such as people, dollars or percent.

Who or what is counted, such as households or people 25 and older.

How it's combined for your shape. See aggregation on each value.

Where it comes from, such as acs5 (the Census Bureau's American Community Survey).

Goes up by one whenever a change at the source alters what the measure counts, so a measure never changes meaning silently.

The kinds of area it's published for.

The years available.

The terms the source publishes it under.

The index's ID.

Its name in plain words.

The current version of its recipe.

The range the score can take, such as 0-100.

The smallest kind of area the score is available for.

A ready-made request by name (GET /presets lists them): its groups and comparisons are added to anything else you ask for. Its suggested area isn't applied for you; send it as around.

Whole topics. One credit for each group in the topic, per place.

Groups of related measures. One credit per group per place.

Single measures. You pay for the group each one belongs to.

Indices. Three credits each per place.

Optional settings. The defaults suit most uses.

Turn every point in your file into an area around it: a radius (type buffer) or a drive-, walk- or bike-time area (type isochrone, 2 credits each to draw). Rows that are already shapes or official areas are used as they are.

The results file: csv (a spreadsheet), geojson (a map file, with each place's outline) or parquet (a compact data file). One row per place, in the order you sent them, with columns id, error and error_detail (if the place failed), area_km2, then each measure by its ID (in a preset's order, if you asked for one) with its margin of error as <id>.moe where it has one (counts from employer records and of places don't), its comparisons as <id>.county, <id>.state and <id>.us and its ranks as <id>.percentile.<county|state|us> when you ask for benchmarks, and a warnings column. Files with addresses also get address_found (the address the geocoder matched) and address_match (exact or close). Default "csv".

The results' column names: ids (measure IDs, such as income.household.median) or labels (plain words with the unit, such as "Median household income (dollars)" and "Median household income (dollars), margin of error"), with the warnings in plain words too. Default "ids".

Which of your columns hold what, when their names aren't the usual ones, such as {"lat": "North", "lon": "East"} or {"address": "Where"}. The keys are id, lat, lon, geoid, wkt, address, street, city, state and zip; null for one stops us guessing it.

Repeat every column of your file in the results, before ours, so each row reads as your row plus its numbers. A name we'd use that your file already has gets ' (Kadrin)' added. Default false.

Optional. An https address on the public internet that we call when the job finishes, with only the job's ID and status; fetch the details with GET /jobs/{id} and your key.

The job's ID. Use it to check progress.

queued (waiting to start), running, succeeded, failed (nothing is charged) or canceled.

Places in your file.

Places finished so far.

Expected cost in credits.

Credits used so far.

Link to download the results once the job has finished (or was canceled part way). Expires after 7 days.

The data release the job uses, fixed when it was accepted.

When the results are deleted.

The name of the file you sent.

The preset you asked for, if any.

When we accepted the job.

When it finished.

If the job failed, what went wrong.

A link to more help on this kind of error, in the docs: https://docs.kadrindata.com/errors#.

A short summary of the error.

The HTTP status code, a standard number for the kind of error, such as 400 (something in the request needs fixing).

• invalid_request: something in the request needs fixing, such as broken JSON, no places, more than 100 places, or an option we don't recognize. The detail names it.
• invalid_geometry: we couldn't read a shape, for example because its outline crosses itself.
• unknown_topic, unknown_variable: we don't recognize that topic or measure ID. Check the catalog.
• topic_unavailable: that topic isn't available yet.
• invalid_key: the API key is missing or wrong.
• insufficient_credits: not enough credits, or the cost is over your Kadrin-Max-Credits limit.
• scope_denied: your key isn't allowed to do this.
• geometry_too_large: a shape is bigger than the limit.
• unavailable_combination: that measure isn't available for that kind of area or those years.
• rate_limited: too many requests too quickly. Wait a moment and try again.
• routing_unavailable: we can't draw drive-time areas right now. Try again shortly.
• not_found: there's nothing with that ID.
• internal_error: something broke on our side. Try again, and contact support with the request_id if it keeps happening.
• address_not_found: in a /jobs file, we couldn't find a row's address. Check it for typos, or give lat and lon instead.

What happened and what to do next.

Which place failed, if the error is about one place.

An ID for this request. Include it if you contact support.