Fair Data Adaptation with Quantile Preservation
Convenience function for returning adapted data
Plotting data before and after adaptation
Compute quantiles generic for the quantile learning step
fairadapt: Fair Data Adaptation with Quantile Preservation
Fair data adaptation (fairadapt)
Fairadapt boostrap wrapper
Fair twin inspection convenience function
Obtaining the graphical causal model (GCM)
Plotting data before and after adaptation
Prediction function for new data from a saved fairadapt object
Prediction function for new data from a saved fairadaptBoot object
Quality of quantile fit statistics
Quantile engine constructor for the quantile learning step
Summarizing fairadapt fit
Summarizing fairadaptBoot fit
Visualize graphical causal model
An implementation of the fair data adaptation with quantile preservation described in Plecko & Meinshausen (JMLR 2020, 21(242), 1-44). The adaptation procedure uses the specified causal graph to pre-process the given training and testing data in such a way to remove the bias caused by the protected attribute. The procedure uses tree ensembles for quantile regression. Instructions for using the methods are further elaborated in the corresponding JSS manuscript, see <doi:10.18637/jss.v110.i04>.