Supervised Learning GMLE Practice Questions (Page 7)

Questions 31–35

1. 31foundation · easy

   A model performs very well on training data but poorly on unseen test data. What is this phenomenon called?

   • ABias-variance tradeoff
   • BUnderfitting
   • CData leakage
   • DOverfitting
   Reveal Solution Discussion Select an answer first
2. 32application · easy

   A robotics company wants to train a robot to navigate a maze. The robot receives a reward for reaching the exit and a penalty for hitting walls. The robot learns by trial and error over many episodes. Which type of machine learning is being used?

   • ASemi-supervised learning
   • BUnsupervised learning
   • CSupervised learning
   • DReinforcement learning
   Reveal Solution Discussion Select an answer first
4. 33expert · hard

   A team is building a model to predict customer churn. They split the data into training (70%) and test (30%) sets. After training, they use the test set repeatedly to tune hyperparameters and select the best model. What is the main problem with this approach?

   • AThe model will underfit because of repeated testing
   • BThe test set should be larger than the training set
   • CThe test set is no longer an unbiased estimate of generalization performance
   • DThe training set is too small
   Reveal Solution Discussion Select an answer first
5. 34application · medium

   A team is building a model to classify images of animals. They have a dataset of 10,000 labeled images. They split the data into training (80%) and test (20%) sets. After training, they evaluate the model on the test set and achieve 95% accuracy. However, when deployed, the model performs poorly on images taken in different lighting conditions. Which practice would most likely improve the model's robustness?

   • AIncrease the size of the test set to 50%
   • BUse data augmentation (e.g., brightness, contrast adjustments) during training
   • CUse a smaller model with fewer parameters
   • DTrain the model for more epochs
   Reveal Solution Discussion Select an answer first
6. 35application · medium

   A startup has a dataset of customer transactions but no labels indicating fraudulent activity. They want to identify unusual patterns that might indicate fraud. Which approach best fits their situation?

   • AUse reinforcement learning to maximize reward for detecting fraud
   • BUse unsupervised learning to cluster transactions and flag outliers for manual review
   • CApply linear regression to predict the probability of fraud
   • DTrain a supervised classification model using transaction amount as the target
   Reveal Solution Discussion Select an answer first