Training and Validation Datasets Quiz

What you'll be asked

1. What is the primary purpose of splitting your dataset into training and validation sets?
2. You are training a classification model and observe that it achieves 99% accuracy on the training data but only 60% accuracy on the validation data. What is the most likely problem?
3. When performing k-fold cross-validation, what is the role of the validation sets?
4. Consider a scenario where you have a highly imbalanced dataset for fraud detection (very few fraud cases vs. many non-fraud cases). You split your data into training and validation sets using a standard random split. What potential issue might arise?
5. You are building a recommendation system. You have a dataset of user-item interactions. You decide to split this data chronologically, using older interactions for training and more recent interactions for validation. What is the primary benefit of this approach over a random split?

Answers and explanations come after you submit.