Address Fraud Detection with Imbalance and Concept Drift Solutions

Q: Address Fraud Detection with Imbalance and Concept Drift Solutions

This question evaluates a data scientist's competence in designing end-to-end machine learning systems for fraud detection, emphasizing challenges such as delayed labels, severe class imbalance, and evolving data distributions (concept drift) in near-real-time scoring.

Q: How do I approach Machine Learning interview questions?

Machine Learning questions require understanding of core concepts and practice. PracHub provides solutions with explanations to help you master machine learning interviews.

Question

End-to-End ML Workflow: Online Payments Fraud Detection

Scenario

You are designing a fraud-detection system for an online payments product that must score transactions in (near) real time. Labels for fraud (e.g., chargebacks) arrive with delays, fraud is rare (severe class imbalance), and fraud patterns evolve over time (concept drift).

Task

Outline the end-to-end ML workflow, covering:

Data collection and labeling
Feature engineering
Model selection and training
Validation and offline evaluation
Deployment and inference
Monitoring and retraining

Additionally, explain how you would handle:

Severe class imbalance
Concept drift

Note: Discuss techniques such as resampling, cost-sensitive learning, ROC-AUC/PR-AUC, sliding windows, and automated retraining triggers.

Address Fraud Detection with Imbalance and Concept Drift Solutions

Overview

End-to-End ML Workflow: Online Payments Fraud Detection

Scenario

Task

Solution

Comments (0)