Tabellab
Data & Analytics Intermediate

Fraud Detection Rules: Velocity Checks, Rule Scoring & Alert Tuning

Build card-not-present fraud rules the way risk teams do: amount-vs-average, new-device and country features, rolling velocity windows with COUNTIFS, weighted risk scores, precision and dollar recall per threshold, analyst capacity, net benefit, a rule performance table and a champion/challenger test, in 8 graded labs. Synthetic data, defensive training only.

8 lessons 8 labs 1.9 h Certificate

The first lesson is free โ€” no account needed.

About this course

A fraud detection rules course that builds the defensive toolkit of a card-not-present fraud team for one fictional online outdoor store, Brightwater Gear, on synthetic transactions in a spreadsheet, for fraud and risk analysts, payments and e-commerce staff, and data analysts moving into risk. You turn checkout attempts into features such as the amount against the card's average, a new-device flag and a country mismatch, count attempts and dollars per card in rolling 24-hour and 1-hour windows with COUNTIFS and SUMIFS, weight five rule hits into one risk score with SUMPRODUCT, back-test thresholds for precision, recall and dollar recall, match alert volume to analyst capacity, find the threshold with the highest net benefit, rate every rule in a performance table and shadow-test a challenger rule set with a two-proportion z-test. Defensive training on synthetic data only. The first lab is free.

Lessons

  1. Transaction Features: Amount vs Card Average, New Device & Country MismatchLab12 minFree
  2. Velocity Checks: Attempts and Amounts per Card in Rolling 24-Hour and 1-Hour WindowsLab15 min Pro
  3. Rule Hits & a Weighted Risk Score: Approve, Review or DeclineLab14 min Pro
  4. Choosing the Alert Threshold: Precision, Recall and Dollar Recall against Labelled FraudLab15 min Pro
  5. Alert Volume vs Analyst Capacity: Staffing, the Lowest Workable Threshold and a Sale-Day BacklogLab14 min Pro
  6. False-Positive Cost vs Fraud Loss Prevented: the Net-Benefit Threshold (with a Chart)Lab14 min Pro
  7. Rule Performance Table: Hit Rate, Precision and Fraud Share per Rule (with a Chart)Lab15 min Pro
  8. Champion vs Challenger: Shadow-Testing a New Rule Set Before It Goes LiveLab15 min Pro