Benchmark Presets:
Active Engine:
RADAR RISK SCORE
C_MULE_8841
98 / 99
BLOCKED
Normal (0-20)
Elevated (21-65)
High Risk (66-99)
Top Feature Attributions (SHAP Local Impact)
"Account exhibits classic mule characteristics: 98% balance drain within 2 hours of receiving funds, asymmetric out/in degree ratio of 16.0, and 4.5x surge over 7-day velocity."
Nearline Redis Feature Store
Pre-populates 1,000 risk vectors into Redis to ensure sub-15ms payment gateway SLA.
C_MULE_8841
MULE HUB
Risk Score:
98.2%
In / Out Deg:
3 / 48
GAT Attn α_ij:
0.9420
PageRank:
0.0085
Identified Mule Account Node
Compromised Source Account
Cash-Out Exit Gateway
[SYSTEM] Connected to Radar live payment pipeline...
[REDIS] Feature store connected. Nearline cache hot.
[PIPELINE] Batch scoring worker online. Target p99 < 15ms.
STREAM VELOCITY:
24 events/sec
TOTAL PROCESSED:
1,482 transactions
BLOCKED MULES / FRAUD:
14 flagged (0.94%)
REDIS CACHE HIT RATE:
97.4% (<0.8ms read)
P99 SERVING LATENCY:
1.12 ms (SLA: <15ms)
Reference (Training) vs Production Distribution
■ Reference (Base)
■ Current (Prod)
CALCULATED POPULATION STABILITY INDEX
0.042
MODEL STABLE
PSI < 0.10: Zero significant covariate drift. Model inference is valid and production-stable.