fraud label

fraud label is the binary target column marking each transaction row as fraudulent or legitimate. Of the 1,744 datasets Datadory catalogs, Kaggle – Credit Card Fraud Detection (ULB) – 284,807 European Card Transactions and Kaggle – IEEE-CIS Fraud Detection – 590K+ Transaction and Identity Tables (Vesta) both document it in their listings.

In this catalog it shows up concretely:

• Kaggle – Credit Card Fraud Detection (ULB) – 284,807 European Card Transactions — One row per card authorization/transaction with a fraud label; bulk download delivery, quality score 8/10, free.
• Kaggle – IEEE-CIS Fraud Detection – 590K+ Transaction and Identity Tables (Vesta) — One row per online payment transaction, optionally enriched with device/identity attributes; bulk download delivery, quality score 9/10, free.

All three Kaggle transaction datasets in this slice ship one, enabling supervised training out of the box.

A field you cannot interpret is a column you cannot join, filter or audit - and misreading one key field corrupts every downstream number.

Kaggle – Credit Card Fraud Detection (ULB) – 284,807 European Card Transactions documents its handling precisely: "One row per card authorization/transaction with a fraud label". When a vendor cannot describe their equivalent field that concretely, expect silent schema drift at integration time.

To verify a source really provides fraud label, work through:

1. Open Kaggle – Credit Card Fraud Detection (ULB) – 284,807 European Card Transactions and compare the claim against the artifact - its own listing reports One row per card authorization/transaction with a fraud label.
2. Open Kaggle – IEEE-CIS Fraud Detection – 590K+ Transaction and Identity Tables (Vesta) and compare the claim against the artifact - its own listing reports One row per online payment transaction, optionally enriched with device/identity attributes.
3. Ask any commercial vendor which cataloged dataset theirs resembles; if they cannot name a comparable public example, treat the claim as unverified.
4. Get update cadence in writing: only 22.6% of the 1,744 cataloged datasets update daily, so "always current" needs evidence.

Where to see it in context: transaction-payment-processing-services data.

Entries that sit next to fraud label in this glossary:

• credit-card-fraud-detection — Credit card fraud detection is the task of separating fraudulent card and online payment transactions from legitimate ones.