Extraction of materials data exceeding the classical Oliver and Pharr data is a complex task. Therefore, different approaches ranging from dimensional analysis to different machine learning algorithms have been used to interpret indentation data. In this study 2D and 3D nanoindentation simulations will be used to train machine learning algorithms. The goal is to predict elasto-plastic parameters and tip radii. Knowledge about tip wear during indentation experiments, especially high throughput experiments, is important for the quality of the obtainable data. The game theory based, model agnostic SHAP (Shapley Additive exPlanations) approach will be used to rank features according to their importance in the only game-theoretical correct way. Thereby light will be shed on the impact of different features, improving the interpretability of the respective model. SHAP is expected to enhance the understanding of machine learning problems in the field of materials mechanics and many other areas as well.