Skip to content

BaseModelBiasChecker

Overview

A model_bias_checker is a component that takes a ML model and runs some checks on that model. These model checks are typically concerned with things like bias and fairness.

Attributes

BaseModelBiasChecker contains no default attributes.

Configuration

BaseModelBiasChecker contains no default or required configuration.

Interface

The following methods are part of BaseModelBiasChecker and should be implemented in any class that inherits from this base class:

check_model_bias

Checks a model for bias. 

def check_model_bias(self, data, model, *args, **kwargs) -> Any

Arguments:

  • data (dict): A dictionary contains the data splits.
  • model (object): The model to check.

Returns:

  • A dictionary of bias metrics and their values.