ACSIncome

ACSIncome is one of the five pre-defined tasks in the Folktables dataset [17], which was recently collected to improve the older and commonly used UCI Adult Income dataset [19]. More specifically, we use the subset of Folktables dataset from the state of California, USA for 2018. The dataset contains a total of 195, 665 data points with 10 features each, where each data points represents an individual. The task is a binary classification to predict whether the individual’s income is above $50, 000. For fairness measures, we use perceived gender (Sex) as the sensitive attribute, which is also one of the 10 input features.