What type of predictions order cases based on the inputs' relationships with the target?

Ranking predictions specifically focus on the order of cases by evaluating how the input variables relate to the target variable. This approach is essential in applications where determining the relative standing of various cases is more important than just predicting particular outcomes. For example, in marketing, a ranking model can segment customers based on their expected purchasing behavior, allowing businesses to prioritize their outreach efforts effectively.

The essence of ranking predictions lies in their capability to produce a sequence that reflects the strength of association between the inputs and the target. In contrast, other types of predictions, such as decision or target predictions, may provide distinct outcomes or classifications without necessarily establishing an order. Estimate predictions also focus on providing numerical values but do not inherently imply an order based on relationships.

This unique characteristic of ranking predictions makes them a valuable tool in contexts where prioritization or ranking is essential for strategic decision-making.