The Partial Dependency tool creates partial dependency plots, as a well as an assessment of each predictor's impact, which is defined as the range of the predicted value of the target across range of each predictor, averaging over the others. A partial dependency plot indicates how the target variable changes across the range of values of a predictor variable. This provides insights into how a predictor influences the target variable, and can be help to determine if the model has "face validity" (i.e., makes intuitive sense) even for models that are "black box" in nature.