Deterministic Synthetic Datasets
Datasets & Feature Distributions
Bundled datasets are generated deterministically from a seed at runtime — inspect class separation and legitimate overlap before writing rules.
ACTIVE PROFILE · SEED #4201
E-Commerce Card-Not-Present (20k)
20,000 synthetic online checkout authorizations (~2.0% fraud) blending card-testing bursts, new-account high-ticket fraud, cross-border stolen cards, and legitimate travelers.
Total Txns
20,000
Actual Fraud
2.00%
400 rows
Total Volume
$2.39M
Fraud Volume
$180,187
Avg Ticket
$450.47 fraud
$112.92 legit
Geo Mismatch
59.3% fraud
8.0% legit overlap
Transaction Amount ($)
Order dollar distribution
Legit %Fraud %
$0–50
$50–150
$150–300
$300–600
$600–1.2k
$1.2k+
Account Age (days)
Avg 24d fraud vs 278d legit
Legit %Fraud %
0–2d
3–7d
8–30d
31–90d
91–365d
365d+
IP Distance (km)
Billing vs IP geolocation gap
Legit %Fraud %
0–50km
50–250km
250–1k
1k–3k
3k–6k
6k+ km
Card Velocity (1h)
Same-card attempts in 60 min
Legit %Fraud %
1
2
3
4–5
6–8
9+
Hour of Day (UTC)
Overnight vs daytime skew
Legit %Fraud %
00–03
04–07
08–11
12–15
16–19
20–23
Composite Velocity Score
0–100 multi-entity velocity index
Legit %Fraud %
0–20
20–40
40–60
60–80
80–100
Fraud Rate by MCC
6051 Crypto / Quasi-Cash1.9% (292/15568)
5411 Grocery0.3% (14/4338)
5732 Electronics100.0% (60/60)
5815 Digital Goods100.0% (20/20)
5691 Apparel100.0% (14/14)
Fraud Rate by AVS Code
Y — Full Match0.4% (67/14971)
U — Unavailable4.5% (223/4919)
N — No Match100.0% (110/110)
Fraud Rate by BIN Country
US0.9% (106/11496)
IN0.3% (26/8236)
DE100.0% (268/268)