KGP Synthetic Customer Behavior Segments for Pawn Lending in New York

New York pawn lender King Gold & Pawn provides a synthetic dataset of 6,643 customer behavior records. Created by gene fur and released on 2026-04-11, it models customer visit frequency, ticket size, collateral preferences, and repayment risk under a consumer stress scenario. The data contains no real individual identities and is designed for research on behavioral segmentation.

Use Cases

• Modeling customer default probability based on the described ticket-to-default correlation of 0.49.
• Analyzing repeat-use customer behavior based on the average annual visit frequency of 4.33.
• Segmenting pawn customers based on described features like visit cadence, ticket size, and collateral preferences.
• Testing retrieval-augmented generation (RAG) systems on distinct customer segments as mentioned in the description.

Strengths

• Contains 6,643 rows of synthetic customer data, providing a substantial sample for analysis.
• Includes specific modeled metrics such as an average annual visit frequency of 4.33 and a ticket-to-default correlation of 0.49.
• Explicitly excludes real customer-level data, addressing privacy concerns for sensitive financial behavior.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Data is synthetic with a realism score of 1.0, which may limit direct applicability to real-world scenarios.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

gene fur via figshare, with mirrors on Zenodo, OpenML, and Kaggle.

Collection Method

Synthetically generated to model pawn customer behavior patterns.

Freshness

Last updated 2026-04-11 00:15:50; freshness should be verified.

Geography

Modeled on operations in New York, including Freeport, Brooklyn, Bronx, and Westchester.