Mathematics Question
1. Empirical Study: Cross-Validation Method 1
Load the mnist.csv dataset. This is exactly the training dataset that you used in HW1, where you fit several knn models to this dataset and found a “best” value of k. In this problem, you will estimate the error of this procedure using cross-validation:
(1) Do 3-fold cross-validation, using this single value of k for the knn model.
Report the estimated error.
(2) Open HW1, where you trained on the same dataset and found the test error. How do these results compare? How would you expect them to compare in general?
Note: Doing this question and the question from HW1 is a little bit silly. It is intended as a bit of a warmup for cross-validation.
DETAILED ASSIGNMENT
