You are planning a dataset whose features have each been rescaled to , so every sample lives in the unit cube .
To make sure a nearest-neighbour model never has to reach far for a neighbour, you want the samples dense enough to cover the cube: partition every axis into equal cells, and require at least one sample in each of the resulting grid cells.
Your storage budget allows at most samples in total.
Use .
What is the largest number of features you can still cover under this budget?
Once you get it right, the reasoning behind the answer shows up here.
Select all that apply.