Zalingo Synthetic Finance, Card-Not-Present Fraud & Chargebacks

Need larger volumes or scheduled refresh? After purchasing this focused sample, message us about enterprise bundles and monthly/weekly/daily subscriptions (CNP only or mixed).

Dataset Features (representative)

• transaction_id, account_id — Synthetic identifiers (non-linkable).
• ts_utc, amount, currency — Event time, value, ISO-4217 code.
• cnp_flag, channel — CNP=1; ecommerce | wallet | mail/phone.
• merchant_id, merchant_country, mcc, category — Merchant context.
• device_fingerprint, ip_country, user_agent — Client hints (synthetic).
• three_ds_result, avs_result, cvv_result — Auth signals (categorical).
• auth_result, decline_reason_code — Approved/Declined + reason.
• velocity_txn_1h / 24h / 7d — Synthetic velocity aggregates.
• distance_km_billing_shipping — Geo-consistency proxy.
• first_time_merchant_flag, recurring_flag, coupon_used — Behavioural context.
• fraud_label (0/1), risk_score_0_1 — Labels + calibrated score.
• chargeback_flag (0/1), chargeback_reason_code — Post-event outcome. (Exact columns may vary slightly; see included preview for this sample’s schema.)

Distribution

• Format: ZIP with Parquet shards (Snappy) + README.
• Volume: 100,000 rows, ~20–30 columns, 1–5 parts.
• Size: ~3–6 MB zipped.
• Schema stability: Consistent across focused CNP bundles; full datasets partitioned by date/merchant/MCC.

Usage

• Fraud/Risk modelling: baselines, uplift vs rule-sets, threshold tuning.
• Authorization optimisation: AVS/3DS policy experiments.
• Scenario testing: velocity, geo-mismatch, first-use patterns.
• MLOps QA: schema/drift checks, dashboard demos.

Coverage

• Geographic: Multi-country synthetic coverage (ISO codes).
• Time Range: Recent multi-year synthetic window with weekly/seasonal patterns.
• PII: None — fully synthetic; not re-identifiable.

License

Who Can Use It

• Risk/Data Science — feature engineering & model iteration.
• Payments/FinOps — auth strategy & loss-rate simulations.
• Product/Analytics — KPI sandboxes w/out sensitive data.

Notes / Disclaimers

• Not real cardholder data. Not for production credit decisions.
• Labels and rates follow synthetic, calibrated distributions and do not represent any specific issuer/acquirer.