Ground Truth Labeling Quiz

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

Quiz covering Feature Engineering

What you'll be asked

  1. What is the primary purpose of 'ground truth' in a supervised machine learning project?
  2. When performing manual labeling for a sentiment analysis task, what is the best practice to ensure high-quality ground truth?
  3. In the context of 'Label Leakage,' why is it dangerous to include a feature that is highly correlated with the target variable?
  4. You are labeling images for a self-driving car project. You notice that your dataset has 99% 'clear road' images and 1% 'obstacle' images. Which labeling strategy should you prioritize?
  5. When constructing ground truth for a temporal sequence prediction model, what is the most significant risk regarding 'look-ahead bias'?

Answers and explanations come after you submit.