Build a Credit-Card Fraud Detector for a Singapore Neobank
Overview
What this challenge is about.
You receive 9 months of anonymized authorization data (around 8 million transactions, around 0.4 percent fraud) plus current rule outcomes. Split temporally and train at least two classifiers (a logistic-regression baseline and a gradient-boosted model such as LightGBM). Calibrate probabilities, sweep operating thresholds, and choose two operating points: one that matches the current rules' fraud capture at a lower false-positive rate, and one that maximizes fraud capture under a hard 1.5 percent false-positive ceiling. Deliver the model, a calibration plot, an operating-point memo, and a deployment plan that addresses concept drift and label delay.
The Brief
What you'll do, and what you'll demonstrate.
Deliver a calibrated fraud-detection model and a deployment plan that beats the current rules on either fraud capture or false-positive rate.
Earning criteria — what you'll demonstrate
- Train classification models under severe class imbalance honestly
- Calibrate model probabilities so thresholds mean what they say
- Translate a model into one or two business-meaningful operating points
- Plan deployment artifacts (monitoring, retraining cadence) before shipping
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Data Scientist
Owning a binary classifier from raw data to a business-defensible threshold is the bread-and-butter project a junior data scientist ships in a fintech risk team.
This challenge sharpens
- classification-modeling
- feature-engineering
- model-evaluation
Machine Learning Engineer
The deployment plan plus calibrated model is the handoff package an MLE turns into a real-time scoring service.
This challenge sharpens
- model-calibration
- python
- model-evaluation