Problem bank

Problem 149 of 333MediumStatisticsP149

Train, validation, test: what does each measure?

  1. Why do you need separate validation and test sets? Why is reporting test accuracy after model selection problematic if you only have train and test?