There are several metrics allowing one to measure bias. We present a few in the following pages. Note that the specific context or model used in your project may mean that these definitions don’t apply. The following definitions will apply to a binary classification problem (supervised learning) where an individual is either classified as 1 (pass) or 0 (fail). This is the most studied case in the literature, as it is relevant for instance for loan applications, recruitment algorithms or academic admission.
An example of how to measure bias in a binary classification problem in recruitment can be found in our notebook, which can be accessed here or downloaded as the following file: