33 Free Fake Datasets by Industry: From Orders and Healthcare to Climate

Most fake data generators start with an empty table and ask you to define every column. That is fine when you know the schema, but slow when you just need a believable orders table, a set of patient visits or a year of weather for a demo. The Data Generator ships with 33 ready-made datasets in 9 groups, and you can change any of them. This guide lists what is in it, which tables join to each other, and how to pick the right one.

Open the Data Generator

What you get

  • **33 ready-made datasets** in 9 groups: General, Healthcare, Pharma, Finance and insurance, Travel and logistics, IT and telecom, Real estate, Marketing, and Climate and weather. A Custom option starts from a blank table.
  • **165 column types** to build or change a dataset, from names, emails and dates to industry types such as diagnosis codes, IBANs and weather readings.
  • **Up to 500,000 rows**, downloaded as CSV, Excel, JSON or Parquet.
  • **A percentage of empty values for each column**, so you can test how your code handles missing data.
  • It runs in your browser. Every value is random and fictional, and none describes a real person or company.

How to use it

  • Change the columns if you like: rename them, change the data type, or set a share of empty values.
  • Set the number of rows, up to 500,000.
  • Choose CSV, Excel, JSON or Parquet and download the file.

To create several related tables at once, use the Linked set tab, which lets you connect columns between datasets.

General: users, orders, products and more

The General group has eight datasets for the common cases:

  • **Users** (1,000 rows): user_id, name, email, phone, age, city, country, signup_date and an active flag.
  • **Orders** (1,000 rows): order_id, customer_id, product_id, sold_by, quantity, unit_price, status and order_date.
  • **Products** (500 rows): sku, name, category, price, stock and rating.
  • **Employees** (500 rows): name, job title, department, salary and hire date.
  • **Transactions** (5,000 rows), **IoT readings** (5,000 rows), **Students** (200 rows) and **Exam marks** (1,000 rows).

These line up at their default sizes. Orders use customer ids from 1 to 1,000, product ids from 1 to 500 and seller ids from 1,001 to 1,500, which match the Users, Products and Employees datasets. Exam marks use the same student ids as Students. So you can join them straight away:

-- DuckDB
SELECT u.country, count(*) AS orders, sum(o.quantity * o.unit_price) AS revenue
FROM read_csv('orders.csv') o
JOIN read_csv('users.csv') u ON o.customer_id = u.user_id
GROUP BY u.country
ORDER BY revenue DESC;

Healthcare

Five datasets: **Patients**, **Hospital visits**, **Lab results**, **Patient vitals** and **Health insurance claims**. Patient ids run from 100001, so Patients, Visits, Lab results and Vitals join on patient_id, and claims also link to visits through encounter_id. Visits carry a diagnosis code and its matching name, a doctor and charges. Lab results have a test, value, unit, reference range and flag, and vitals have heart rate, blood pressure, temperature and oxygen saturation.

Names of drugs, diagnoses and airports are real public terms so the data looks right, but no value describes a real patient. See fake patient data for healthcare testing for a worked example.

Pharma

**Prescriptions** (drug, class, strength, form, dosing, pharmacy and status), **Drug inventory** (stock, batch number, expiry and unit cost) and **Clinical trial subjects** (trial, phase, arm and adverse events with severity). Prescriptions use the same patient ids as the Healthcare datasets. More in pharma sample data.

Finance and insurance

Seven datasets: **Bank accounts** (IBAN, balance, credit score), **Loans**, **Insurance claims**, **Trade blotter** (symbol, side, price, counterparty, fees), **Bank transactions**, **Booking payments** and **General ledger**. Bank transactions and the ledger carry card numbers, IBANs, vendors and approver names, so they are a good way to practise masking. See fake bank, loan and insurance claims data and masking bank transactions.

Travel, IT and telecom, real estate and marketing

  • **Travel and logistics:** Flight bookings (route, fare, class) and Shipments (tracking number, carrier, weight, status). Booking payments is designed to link to Flight bookings through booking_ref. See flight and shipment sample data.
  • **Marketing:** Ad campaign performance, where impressions, clicks, spend and conversions add up (clicks never exceed impressions), and CRM contacts with safe example.com emails. See sample marketing campaign data.

Climate and weather

The newest group has three datasets that fit together:

  • **Weather stations** (30 rows): station_id, city, country, latitude, longitude, elevation, climate zone and installation date. The cities range from Singapore to Ulaanbaatar.
  • **Daily weather** (5,000 rows): one row per station per day, with minimum, mean and maximum temperature, humidity, precipitation, wind speed and direction, air pressure, UV index and conditions. You can set the number of stations (1 to 30) and the first day.
  • **Climate indicators** (780 rows, monthly from 1960): CO2, temperature anomaly, sea level and Arctic sea ice. You can change the first year.

The columns of a row agree with each other. In a test run with 366 days for every station, London averaged about 5 C in January and 18 C in July, while Sydney averaged about 23 C in January and 12 C in July, so the seasons flip in the southern hemisphere. Minimum is always below mean and mean below maximum. Days with rain, snow or a storm are exactly the days with precipitation. Air pressure falls with height: about 1009 hPa at London, 833 at Denver and 645 at La Paz. In the indicators, CO2 started near 314 ppm in January 1960 and reached about 421 ppm in May 2023, and it is higher in May than in September.

Remember what this is. The values are plausible but synthetic, so they are not measurements of any real day or place. They are for demos, tests, dashboards and teaching, not for climate research. Join the two tables on station_id:

SELECT s.city, round(avg(d.temp_mean_c), 1) AS avg_temp_c, sum(d.precip_mm) AS total_rain_mm
FROM read_csv('daily_weather.csv') d
JOIN read_csv('weather_stations.csv') s USING (station_id)
GROUP BY s.city
ORDER BY avg_temp_c DESC;

Which dataset for which job

  • **Practise SQL joins:** Orders with Users and Products, or Patients with Hospital visits. See practise SQL on a CSV.
  • **Practise masking:** Bank transactions, General ledger or Patients. See data masking.
  • **Demo a dashboard:** Ad campaign performance, Property listings or Daily weather.
  • **Time series:** IoT readings or Daily weather. **Maps:** Weather stations have latitude and longitude.

Common mistakes

  • **Generating tables separately and then failing to join them.** Use the default sizes, which line up, or the Linked set tab.
  • **Testing performance with too few rows.** A thousand rows tells you nothing about a million. Generate up to 500,000 rows and export Parquet to keep the file small.
  • **Treating generated data as evidence.** It is random, so it shows how your code behaves, not what the real world looks like.
  • **Forgetting missing values.** Real data has gaps. Set a percentage of empty values on the columns that need them.
  • **Mixing it up with real data in a shared system.** Name test files clearly so fake data is never mistaken for production data.

For how the generator works in general, see the free mock data generator guide, and for building a dataset from your own table definition, see generate test data from a CREATE TABLE. To keep tables consistent across datasets, read test data with foreign keys.

Open the Data Generator

Related tool pages

  • Fake Data Generator: Generate realistic fake data online: names, emails, addresses, dates and more, up to 500,000 rows, as CSV, JSON, Parquet or Excel. Free.
  • Test Data Generator: Generate test data for databases and apps: names, emails, addresses, dates and numbers, up to 500,000 rows, as CSV, JSON, Parquet or Excel. Free.
  • Student Data Generator: Fake Student Records: Generate fake student data for class projects: IDs, names, courses, marks and attendance. Download CSV, Excel, JSON or Parquet. Free.
  • Student Marks Dataset Generator: Create a student marks dataset by student, subject and exam for SQL, Excel pivot tables and charts. Free, in your browser.
  • Sample Data for SQL Practice: Generate sample tables for SQL practice (students, orders, employees), download CSV and query it in your browser. Free.
  • Dummy Data for Excel Practice: Generate dummy sales, student or employee data and download Excel to practise filters, pivot tables and VLOOKUP. Free.

Also try

How to use the Data Generator: step by step

  1. Open the Data Generator and choose a group and a dataset.
  2. Adjust the columns, empty values and row count.
  3. Choose CSV, Excel, JSON or Parquet.
  4. Download the file and load it into your tool.

When to use the Data Generator

  • Generate test data for a database or an app
  • Practise SQL joins on realistic tables
  • Create sample data for a dashboard demo
  • Make a file to practise masking

Example

Input: Dataset: Customers, 3 rows

Output: id,name,email 1,Maya Patel,maya.patel@example.com 2,Leo Fischer,leo.fischer@example.com

Frequently asked questions

What kinds of test data can I generate?

There are 33 ready-made datasets in 9 groups: General (users, orders, products, employees and more), Healthcare, Pharma, Finance and insurance, Travel and logistics, IT and telecom, Real estate, Marketing, and Climate and weather. You can also build a custom table.

How many rows can the Data Generator create?

Up to 500,000 rows, downloaded as CSV, Excel, JSON or Parquet.

Do the generated tables join to each other?

Many do at their default sizes. Orders match Users, Products and Employees by id. Patients, Hospital visits, Lab results, Vitals and Prescriptions share patient ids. Weather stations and Daily weather join on station_id. The Linked set tab connects other datasets.

Is the climate data real?

No. It is plausible but synthetic: seasons, rain, pressure and trends look realistic, but the values are not measurements of any real day or place. Do not use them for research.

Is the generated data private?

Yes. The data is generated in your browser, and every value is random and fictional, so none describes a real person or company.

Tip: Keep the default row counts when you want tables that join. Change them only when you also change the id ranges, or use the Linked set tab.