<p>This paper explores applicability of two explainable artificial intelligence (XAI) techniques, i.e. local interpretable model-agnostic explanations (LIME) and partial dependence plots (PDP) for parametric analysis of a drilling operation with thrust force, torque and flank wear as the three responses, aiming in enhancing process efficiency through data-driven optimization. Based on a past experimental dataset, three supervised machine learning (ML) algorithms, i.e. random forest (RF), extreme gradient boosting (XGBoost) and adaptive boosting (AdaBoost) are first adopted to frame the prediction models with drill diameter, spindle speed and feed rate as the input parameters. Considering several statistical metrics, the best-fit model having the maximum prediction accuracy is chosen for each of the responses. To address the interpretability gap of the complex ML models, this paper integrates LIME for instance-specific reasoning and PDP to understand effects of the drilling parameters on the responses at a global level. It is revealed that feed rate plays a dominant role in influencing all the responses. Drill diameter has nonlinear impacts on thrust force and flank wear, especially beyond a critical threshold, and spindle speed has only moderate effects, slightly reducing thrust force and flank wear. These findings are consistent with the established machining principles and validate the model outputs. By combining accurate prediction with explainability, it provides a practical and transparent framework for drilling process optimization, aiding in parameter selection, tool life improvement and operational decision making.</p>