AIExplainer

What is demographic parity?

A fairness metric in machine learning that ensures equal outcomes for different demographic groups

Demographic parity is a measure used to evaluate the fairness of a machine learning model by checking if it produces similar outcomes for different demographic groups, such as age, gender, or ethnicity. The goal is to ensure that the model does not discriminate against any particular group.

Demographic parity is like ensuring that a referee in a sports game applies the same rules to all players, regardless of their background or characteristics. Just as the referee should be fair and impartial, a machine learning model should produce fair and unbiased outcomes for all demographic groups.

For example, a company using a machine learning model to screen job applicants might use demographic parity to ensure that the model is not biased against certain age groups or ethnicities. If the model is found to be biased, the company can take steps to adjust the model and ensure that it is fair and unbiased.

Demographic parity is used in machine learning to evaluate the fairness of models, particularly in applications where bias can have significant consequences, such as hiring, lending, or law enforcement. It is often used in conjunction with other fairness metrics to ensure that models are fair and unbiased.

One common misconception about demographic parity is that it requires equal outcomes for all groups, regardless of their qualifications or characteristics. However, demographic parity only requires that the model produces similar outcomes for groups with similar qualifications or characteristics.

The concept of demographic parity has been around for several decades, but it has gained significant attention in recent years with the increasing use of machine learning models in high-stakes applications.

statistical parity fairness metric bias metric

Three products for different needs — explore what’s relevant to you.