Analysis of Complex Models Using XAI and the Development of an Intelligent Agent for Effective Customer Segmentation in Banking

This thesis focuses on the analysis of complex machine learning models using explainability methods (XAI) and the development of an intelligent agent that serves as an interactive interface to enhance the efficiency of customer segmentation in banking. The theoretical section introduces the fundamentals of machine learning, the principles of customer segmentation, and an overview of commonly used clustering methods, with special emphasis on model explainability techniques that enable a deeper understanding of their internal logic and decision-making processes. The practical section concentrates on a detailed analysis of selected clustering algorithms using modern XAI methods, facilitating the interpretation and comprehension of cluster formation. Subsequently, an architecture for an intelligent agent is proposed and implemented, integrating analytical modules and providing users with an intuitive interface for working with segmentation results. Experimental validation on real data from the banking sector demonstrates the practical applicability of the proposed methodologies and their ability to optimize customer segmentation processes. The results of this work contribute to enhancing the transparency of complex machine learning models and offer an innovative approach to support more informed decision-making in banking.