When building a machine learning model, what’s the purpose of a test dataset?
- To generate training labels
- To improve model performance during training
- To evaluate the model’s performance on new, unseen data
- To increase the size of the training dataset