Support for specifying a range during conditional sampling about sdv

[npatki]

Workaround

For anyone blocked by this, you can use the code snippet below. This code will sample a lot of rows (unbounded) and then filter out afterwards to a specific range.

# TODO: input the conditions you need
COL_NAME = 'my_column_name'
LOW_RANGE = 18.0 # minimum possible value in range
HIGH_RANGE = 100.0 # maximum possible value in range

# Request more rows than you need. Maybe 1,000 if you need 100 true rows.
synthetic_data = synthesizer.sample(1000)

# Filter out rows to within the range
filtered_synthetic_data = synthetic_data[synthetic_data[(synthetic_data[COL_NAME] >= LOW_RANGE) & (synthetic_data[COL_NAME] <= HIGH_RANGE)]

[adib0073]

Thanks @npatki for sharing the workaround code. Can such conditions be defined even before generating the samples? I think it would be better to have something like generate with conditions (different from generating with constraints) to avoid unnecessary computation time in generating and then filtering based on conditions.

[npatki]

Hi @adib0073, unfortunately I cannot think of a good workaround that would allow you to do so right now.

However in the future, when the team adds an actual feature to enable range-based conditional sampling, that is exactly how I envision it working.