Data Drift Detection Quiz

5 questions Pass: 70% +25 pts (50 with Pro)

Quiz covering Model Monitoring

What you'll be asked

  1. What is the primary purpose of monitoring data drift in a machine learning production system?
  2. Which of the following statistical tests is commonly used to detect data drift for numerical features by comparing two probability distributions?
  3. If you observe a significant drop in model accuracy but no significant change in the input data distribution, what is the most likely cause?
  4. When monitoring for data drift, why is it important to define a 'reference dataset'?
  5. You are monitoring a high-cardinality categorical feature. Which method is mathematically robust for detecting drift by measuring the difference between the observed frequencies and the expected frequencies?

Answers and explanations come after you submit.