Add LogitScaler transformer

[frances-h]

Problem Description

Currently, the ScalarRange, ScalarInequality, Positive, and Negative constraints all scale the data using log-based transformers. Since the SDV already enforces min/max values by simpler means, these constraints can be deprecated. However, it would be helpful to move the scaling logic into the RDT library, so that it may be used independently of the constraints in the future.

Expected behavior

Create a new RDT called LogitScaler. This RDT transforms the data by applying a logit function. Its functionality is equivalent to the ScalarRange constraint.

Parameters

• missing_value_replacement (object): Same as FloatFormatter
• missing_value_generation (str or None): Same as FloatFormatter
• min_value (float): The min value for the logit function. Defaults to 0.
• max_value (float): The max value for the logit function. Defaults to 1.0.
• learn_rounding_scheme (bool): Same as FloatFormatter

Implementation Notes

• Apply similar logic that is used in ScalarRange
• During fit:
  • Learn needed information about missing values and rounding scheme.
  • Validate that we can actually apply the logit function to the data.
• During transform:
  • Fill in missing values
  • Transform the data
    • Use the logit function to transform the data from range [min_value, max_value] into range (-inf, +inf)
  • If there are any issues with taking the logit, raise a descriptive error explaining what wen wrong:
    • Example: Error: Unable to apply the logit function to column 'output-value' due to an out-of-range value (101.0).

Intended usage: A user would call update_transformers to replace the FloatFormatter with the LogitScaler for any columns that would benefit from this.

synthesizer.auto_assign_transformers(data)
synthesizer.update_transformers({
  'output-value': LogitScaler(min_value=0.0, max_value=100.0)
})
synthesizer.fit(data)