Fraud Detection & Transaction Monitoring

Total Fraud Cases

556

9.3% of all transactions

High Risk Txns

Score ≥ 60 · Immediate review

Fraud Amount

₹12.3M

Total flagged value

Safe Transactions

4,973

Low risk · Score < 30

Monthly Fraud Trend

Fraud cases detected per month (Jan–Dec 2024)

Risk Score Distribution

Transactions by risk category

Fraud by Category

Which merchant categories trigger most fraud flags

Fraud Rule Triggers

Which rules fire most often

SQL

Core Detection Queries

Rule 1 · High Amount Flag

SELECT transaction_id, amount,
  CASE
    WHEN amount > 10000 THEN 'HIGH_AMOUNT'
    ELSE 'NORMAL'
  END AS amount_flag
FROM transactions;

Rule 2 · Location Mismatch

SELECT transaction_id,
  home_location, transaction_location,
  CASE
    WHEN home_location != transaction_location
    THEN 1 ELSE 0
  END AS location_mismatch
FROM transactions;

Rule 3 · Risk Scoring (CTE)

WITH base AS (
  SELECT *,
    CASE WHEN amount > 10000       THEN 40 ELSE 0 END s1,
    CASE WHEN home_loc != txn_loc  THEN 30 ELSE 0 END s2,
    CASE WHEN is_international=1   THEN 20 ELSE 0 END s3
  FROM transactions
)
SELECT transaction_id,
  s1+s2+s3 AS risk_score
FROM base;

Rule 4 · Final Output Table

SELECT transaction_id,
  risk_score,
  CASE
    WHEN risk_score >= 60 THEN 'HIGH'
    WHEN risk_score >= 30 THEN 'MEDIUM'
    ELSE 'LOW'
  END AS risk_category,
  fraud_flag
FROM scored
ORDER BY risk_score DESC;