Closed popescu-v closed 1 month ago
Maybe just implement a helper function sort_dataset
that takes a dictionary file or domain and sorts all tables by their key. The signature would be similar to that of train_predictor
.
The automatic sort for the sklearn
module was implemented in the 10.1.1
with the internal_sort
(defaults to True
). The parameter proposed for core.api
seems irrelevant.
When we will expose the Dataset
we will implement a sort
function for a multitable dataset. So, this feature will be not be necessary in core
and thus is obsolete.
Implementation ideas
sort
totrain_predictor
,deploy_model
,train_recoder
,evaluate_predictor
(defaultFalse
)sklearn
estimators to just use the newsort
option ofcore
functions and eliminate calls tocore.sort_data_table