Exploring Explainable AI And Data Visualization In Data Science For Improved Business Insights

Abstract

The increasing use of data-driven models in business analytics has amplified concerns regarding the transparency and interpretability of predictive systems. In response to these concerns, explainable artificial intelligence (XAI) has emerged as a critical approach for enhancing trust and understanding in machine learning–based decision support. This study investigates the integration of explainable artificial intelligence and data visualization within a data science framework to generate interpretable and actionable business insights, using customer churn prediction as an empirical context. A supervised learning model was developed to predict churn outcomes based on customer demographic, service, contractual, and billing attributes. To support interpretability, both model-intrinsic and post-hoc explainability techniques were applied, enabling examination of feature influence at global and local levels. Visual analytics were employed to translate explainability outputs into intuitive representations, facilitating interpretation across different analytical perspectives. The results demonstrate that churn behavior is influenced by a combination of service configuration, contractual commitment, and customer engagement characteristics, while explainability techniques provide clarity on how these factors contribute to individual predictions. The study highlights the value of combining interpretable modeling, explainable AI methods, and visualization to enhance the practical usability of data science models in business contexts. By emphasizing transparency alongside prediction, the proposed approach supports informed decision-making and contributes to the development of trustworthy analytics systems.