Dataset with experimental results for the "Frank" hybrid decision-making system, with simulated users.

Features: - CA. Co-evolutionary Accuracy. Accuracy reached by the user aided by Frank. Ideal value: 1 - CD. Co-evolutionary Discrimination. Discrimination reached by the user aided by Frank. Ideal value: 0 - UC. Unfair Couples. Number of couples of identical records (save for the sensitive attributes) with a different assigned label. Ideal value: 0 - MA. Model accuracy on a separate dataset. Ideal value: 1 - MD. Model discrimination on a separate dataset. Ideal value: 0 - Dataset: Dataset used for the experiment - Mode: The mode of the HDM system. Possible values: none (i.e., user alone), Rule, Past, Group, Skept, Frank (i.e., everything enable).. Refer to the Frank paper for an explanation - Expert: The kind of simulated user. The Ground Truther, Absent-Minded, and Coin-Tosser choose the dataset's ground truth, respectively, with a probability of 100%, 75%, and 50%. Naive Bayes and KNN are model-based. - Believe: Believe mode of the user, for Frank's suggestion in "Skept" and "Frank" mode. Possible values: (Always) Accept, (Always) Decline, Random (choice), XAI (i.e., request explanation, then choose).