Bank and card transactions/Synthetic
Fraud-labeled transactions
A labeled transaction set for fraud models. Accounts, merchants, and amounts are synthetic. Each row has a fraud label and a split, so the same specification can be used to train and to test. There is no real cardholder data and no real payment credentials.
- Volume
- 5,000,000transactions
- Formats
- CSV · Parquet · JSONL
- Label
- Synthetic
Purchase
Sold on Datarade. This site does not take payment.
Placeholder — listing URL not set
Schema
Column preview for the launch specification. Names and types describe the file. They are not a sample of rows.
- transaction_idstring
Identifier for one transaction.
- account_idstring
Synthetic account identifier. Not a real account.
- event_timetimestamp
When the transaction occurred.
- amountdecimal
Transaction amount.
- currencystring
ISO currency code.
- mccstring
Merchant category code.
- channelstring
How the payment was made, such as card-present or online.
- is_fraudboolean
Fraud label for training and testing.
- fraud_typestring
Fraud category when is_fraud is true. Otherwise empty.
- splitstring
train or test.
Use
- Train a card and bank fraud classifier
- Test a fraud model on a labeled holdout
Generation
Synthetic accounts and transactions. Fraud labels are part of the row. No real cardholder data.
- Every dataset is 100% synthetic. It is not copied from real customers, cardholders, or patients.
- Each product is labeled as synthetic in the title, the description, and the file metadata.
- Delivery includes a generation note: the methods and algorithms used, the schema, and known limitations.