What visualizes the relationship between the target response and a set of input features?

Partial Dependence Plots (PDP) are specialized visual tools designed to illustrate the relationship between a target response and one or more input features while accounting for the effects of other features in the model. These plots allow observers to understand how the predicted outcome behaves as the value of an input feature varies, which can be critical in interpreting complex models like those found in machine learning and AI.

By presenting the marginal effect of feature values on the target response, PDPs help researchers and practitioners visualize the nature of that relationship—whether it is linear, monotonic, or more complex. This insight is invaluable for not only interpreting model predictions but also for refining feature engineering and model selection.

The other options serve different purposes. Feature Interaction Plots look at the combined effects of two or more features and how they interact with one another in influencing the target, which is a more specific insight. Response Curve Maps generally visualize how a response variable changes with respect to predictor variables in a continuous manner, but may not specifically isolate the effect of one input feature as clearly as PDPs. Correlation Heatmaps depict the strength and direction of relationships among multiple variables but do not specifically illustrate the effect of features on a target response in the way that PDPs do.