Roadmap: parsnip support for sparse tibbles #1125
Description
What we need:
-
fit()to take sparse tibbles as data Allow sparse tibbles infit()andfit_xy()#1165 -
fit()to take {Matrix} sparse matrix as data letfit_xy()take dgCMatrix input #1121- turn them into sparse tibbles early
-
predict()to take sparse tibbles as data Make sure sparse matrices can be used withpredict()#1167 -
predict()to take {Matrix} sparse matrix as data Make sure sparse matrices can be used withpredict()#1167- turn them into sparse tibbles early
- look into if we document which engines are sparse friendly document sparse data usage in parsnip #1171
- special cases for some model types make sure xgboost works with sparse data #1173
- {xgboost} with
xgboost::xgb.DMatrix()andxgb_train()(only special case engine that allows sparse data)
- {xgboost} with
- Better error if sparse matrix is used with
fit()Make sure all sparse data errors look nice #1174 - Make sure all error messages are piped through correctly Make sure all sparse data errors look nice #1174
I think we could use a option() of some kind to unit test that the data passed is passed around in a way that keeps the sparsity.
Adding all of this will give us
- standalone usage of sparse matrices in {parsnip}
- everything it needs to be able to work with the rest of {tidymodels} in regards to sparse tibbles