Order the output of get_transformers() based on the metadata

[npatki]

Problem Description

In the new SDV version 1.0, I am able to inspect the transformers that will be used if I run:

synthesizer.auto_assign_transformers(real_data)
synthesizer.get_transformers()

This returns a dictionary mapping each column name to the transformer object.

{
  'age': FloatFormatter(),
  'weight': FloatFormatter(),
  ...
}

Expected behavior

In the returned dictionary from get_transformers(), I observed that the keys were out of order.

It would be nice if the keys were in the same order as the metadata, as the metadata is considered the ground truth. This will make it easier for me to inspect the transformers.

[npatki]

Replication

from sdv.datasets.demo import download_demo
from sdv.single_table import GaussianCopulaSynthesizer

# metadata columns are in the order of: start_date, end_date, address
data, metadata = download_demo(
    modality='single_table',
    dataset_name='student_placements_pii'
)

synthesizer = GaussianCopulaSynthesizer(metadata)
synthesizer.auto_assign_transformers(data)
synthesizer.get_transformers()

This prints out the dictionary in a different order

{
  'experience_years': FloatFormatter(learn_rounding_scheme=True, enforce_min_max_values=True),
 'second_perc': FloatFormatter(learn_rounding_scheme=True, enforce_min_max_values=True),
 'duration': LabelEncoder(add_noise=True),
  ...
}