Synthetic Card Transactions with 5 Fraud Patterns, 500K Records

500,000 realistic synthetic card transactions designed for fraud detection tasks. The description indicates the data includes five distinct fraud patterns and has undergone production-grade feature engineering. The dataset's author, organization, and license are unknown.

Use Cases

• Training fraud detection classifiers based on the five described fraud patterns.
• Benchmarking anomaly detection algorithms on realistic synthetic transaction data.
• Developing and testing feature engineering pipelines for financial transaction data.

Strengths

• 500,000 rows of data, providing a substantial volume for model training.
• Includes five distinct fraud patterns, offering a structured testbed for detection algorithms.
• Description mentions production-grade feature engineering, suggesting practical relevance.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is known but other scale details like file size and formats are unknown.
• Data may reflect bias inherent to its synthetic generation method, which is unspecified.

Provenance

Source

Kaggle

Collection Method

Synthetic generation; specific method unknown.