An actionable set of changes a person can undertake in order to improve their outcome.
A predictive model has recourse if there exists an intervention for every possible input that would change the prediction of the model. For example, a model that makes predictions solely based on race, age or sexual orientation does not allow recourse as there are no actions a person could take that would change the output.
Such a system would be widely viewed as discriminatory.
Proving that a machine learning model ensures recourse is an active area of research.