Prescriptive analytics is the branch of data analysis that recommends what action to take, given an objective and a set of constraints. It is usually described as the last of four stages: descriptive analytics reports what happened, diagnostic analytics explains why, predictive analytics estimates what is likely to happen, and prescriptive analytics proposes the best response. It combines forecasts with mathematical optimisation, simulation methods such as Monte Carlo simulation, business rules and machine learning.
In financial services, typical uses include dynamic pricing of loans and deposits, setting and adjusting customer limits as part of credit limit management, allocating cash and collateral in treasury, routing payments, and choosing how to respond to a suspected fraud, for example whether to block, challenge or approve a transaction. Results depend on the quality of the underlying data and models. Recommendations that affect customers are subject to model risk management and, in some jurisdictions, to rules on automated decision-making.
Ref 1300000412