aif_dataset: AIF360 dataset

In aif360: Help Detect and Mitigate Bias in Machine Learning Models

Description

Function to create AIF compatible dataset.

Usage

aif_dataset(data_path,  favor_label, unfavor_label,
                     unprivileged_protected_attribute,
                     privileged_protected_attribute,
                     target_column, protected_attribute)

Arguments

data_path	Path to the input CSV file or a R dataframe.
favor_label	Label value which is considered favorable (i.e. “positive”).
unfavor_label	Label value which is considered unfavorable (i.e. “negative”).
unprivileged_protected_attribute	A unprotected attribute value which is considered privileged from a fairness perspective.
privileged_protected_attribute	A protected attribute value which is considered privileged from a fairness perspective.
target_column	Name describing the label.
protected_attribute	A feature for which fairness is desired.

See Also

More about AIF binary dataset.

Examples

load_aif360_lib()
# Input dataset
data <- data.frame("feat" = c(0,0,1,1,1,1,0,1,1,0), "label" = c(1,0,0,1,0,0,1,0,1,1))
# Create aif compatible input dataset
act <- aif360::aif_dataset(data_path = data,  favor_label=0, unfavor_label=1,
                            unprivileged_protected_attribute=0,
                            privileged_protected_attribute=1,
                            target_column="label", protected_attribute="feat")

aif360 documentation built on July 1, 2020, 5:34 p.m.