• 4.0 Industry
• AI
• artificial intelligence
• BoCSoR
• coffe roasting
• counterfactual
• explainable artificial intelligence
• feature importance
• machine learning
• product quality
• XAI

Industry 4.0 represents a new era of production characterized by the interconnection of devices and the digitization of processes. AI plays a crucial role in this context, enabling intelligent automation, predictive maintenance, and the optimization of production processes. The industrial dataset used is derived from sensors installed on a coffee roasting machine and contains detailed information about the production process, including temperature, humidity. In this thesis, the focus is not so much on predicting coffee quality based on data, but rather on the application of XAI techniques to understand the input-output relationships in the production process in an interpretable manner. Explainable AI (XAI) techniques emerge as crucial tools for making artificial intelligence models transparent. In particular, counterfactual-based explanations prove effective in revealing input-output relationships in complex models. In the present research, the XAI technique called BoCSoR is employed to analyze both the regression and classification tasks of coffee quality. BoCSoR provides an interpretable and transparent approach to model analysis, allowing for a better understanding of the variables influencing coffee quality. For the regression task, BoCSoR has been parameterized to accurately identify and evaluate relevant features impacting the quality of the coffee produced. This approach enables a deeper understanding of the key factors influencing the final outcome. The analysis of coffee quality in the context of Industry 4.0 represents a significant application of XAI techniques. The use of BoCSoR to explore input-output relationships in AI models offers an interpretable and transparent approach, enabling more accurate optimization of production processes and higher quality of the final product.