This chapter introduces several techniques for global exploration and explanation of a model’s predictions for a set of instances. While combining various techniques for dataset-level explanation can provide additional insights, it is worth remembering that the techniques are, indeed, different and their suitability may depend on the problem at hand. The key element of dataset-level explanatory techniques is the set of observations on which the exploration is performed. Therefore, sometimes these techniques are called dataset-level exploration. In the case of model-performance assessment, it is natural to evaluate it using an independent testing dataset to minimize the risk of overfitting. However, PD profiles or accumulated-local (AL) profiles can be constructed for both the training and testing datasets. The random forest model suggested a U-shaped relationship between the construction year and the price of an apartment, which was missed by the simple linear-regression model.