Add sampling methods to PARSynthesizer

[amontanez24]

Problem Description

As a user, it would be helpful to have a clear understanding of how to sample sequential data whether I have to provide the context or not.

Acceptance criteria

Add the following methods to the PARSynthesizer

• sample
  • Parameters:
    • (required) num_sequences(int): The number of sequences to sample
    • sequence_length(int): An integer that describes the length of every sequence
      • (default) None: The model will determine the length of every sequence
    • randomize_samples (bool):
      • True: Every sample() call will have randomly generated data that is different
      • (default) False: Every sample() call has the same data generated
• sample_sequential_columns
  • Parameters:
    • (required) context_columns (pandas.DataFrame): A DataFrame containing the rows for the unchanging context
    • sequence_length (int): An integer that describes the length of every sequence
      • (default) None: The model will determine the length of every sequence
    • randomize_samples(bool):
      • True: Every sample() call will have randomly generated data that is different
      • (default) False: Every sample() call has the same data generated

Expected behavior

>>> synthesizer.sample(5, sequence_length=10, randomize_samples=True)
>>> context = pd.DataFrame({'name': ['sam', 'sam', 'sam']})
>>> synthesizer.sample_sequential_columns(context, 5, True)

Additional context

• The current PAR model only has one sampling method with a bunch of different arguments. The functionality in that method can essentially be split up into the two methods above.

[npatki]

When doing this change, we can also fix #1052: In sample_sequential_columns the columns provided as context_columns can be in any order. (We can simply reorder them if as needed.)