benchmarking thresholding for when model starts benefitting for using sparse data over dense data.

[EmilHvitfeldt]

Basically, at what sparsity level does having sparse data improve things.

e.i. should a dataset with 10% be converted to a dcgmatrix or stay a matrix?

there are kinda two scenarios. you have sparse tibbles, or dense tibble with a lot of zeroes

[EmilHvitfeldt]

This was done in https://github.com/tidymodels/benchmark-sparsity-threshold.

We will use the results of the analysis in tidymodels/workflows#271