Appendix for "Identifying Imbalance Thresholds in Input Data to Achieve Desired Levels of Algorithmic Fairness"

In this document we provide three appendixes for the journal article “Identifying Imbalance Thresholds in Input Data to Achieve Desired Levels of Algorithmic Fairness”.

In Appendix A we show predictors and targets that we took into account for each dataset employed in our study.

In Appendix B we describe the configurations of the thresholds that we defined during the procedure of Identification of Risk Thresholds. 

In Appendix C, for each combination of balance-unfairness-algorithm we report the best thresholds selected by accuracy, the configuration they correspond to (among the 5 options described in Appendix B), and all the evaluation metrics related to those thresholds.