You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As a user, it would be great if synthesized data matched the same numerical format (decimal places) as the fitted data.
The constraints have their reverse_transform after the HyperTransformer and sometimes ruin the formatting. As opposed to adding this formatting check in every constraint, we can pull out this logic into formatting classes that the DataProcessor can use.
Expected behavior
Create a NumericalFormatter class
Add a learn_format method that learns the required information from a column.
Add a format_data method that applies the learned format to a passed in column.
Additional context
We can decide whether to use this class in the constraints or at a higher level
Problem Description
As a user, it would be great if synthesized data matched the same numerical format (decimal places) as the fitted data.
The constraints have their
reverse_transform
after theHyperTransformer
and sometimes ruin the formatting. As opposed to adding this formatting check in every constraint, we can pull out this logic into formatting classes that theDataProcessor
can use.Expected behavior
NumericalFormatter
classlearn_format
method that learns the required information from a column.format_data
method that applies the learned format to a passed in column.Additional context
learn_format
method can basically copy these lines from theFloatFormatter.fit
https://github.com/sdv-dev/RDT/blob/8d2894ce30635902ccbb5bc9b1e8bc60f388db78/rdt/transformers/numerical.py#L156-L163
format_data
method can basically copy these lines from theFloatFormatter.reverse_transform
https://github.com/sdv-dev/RDT/blob/master/rdt/transformers/numerical.py#L204-L217
The text was updated successfully, but these errors were encountered: