WIP: For augmentation, cast dataframe to profiled types
Currently pandas is selecting the dtype automatically, which might not match the profiled types exactly (e.g. profiling accepts a small quantity of unclean types, and knows more types like temporal). Only the index columns are casted (in set_data_index()
, for the join to work).
This will make the aggregations work as expected given the profiled data types.
See also !13 (closed) (might want to close?)