<![CDATA[Digital transformation in the automotive industry requires organizations to effectively integrate and analyze transactional data to support strategic decision-making. However, fragmented and unstructured sales data often limit comprehensive market segmentation and performance evaluation. This study aims to design a data mart to support the development of Predictive Key Performance Indicators (KPIs) for automotive market segmentation analysis using TYT Delivery Order (DO) data from 2020–2023. The research methodology includes the design of a data warehouse based on a star schema model, implementation of the Extract–Transform–Load (ETL) process, calculation of Recency, Frequency, and Monetary (RFM) metrics, customer segmentation using the K-Means clustering algorithm, and the development of Machine Learning models to generate predictive KPIs such as Purchase Probability, Churn Risk Score, and Predicted Customer Lifetime Value (CLV). The results indicate that the proposed data mart successfully integrates historical transaction data, segmentation outputs, and predictive scores into a unified analytical structure optimized for Business Intelligence dashboards. The analysis reveals significant differences in revenue contribution across customer clusters and demonstrates the effectiveness of predictive models in identifying high-potential customers and those at risk of churn. In conclusion, the integration of a data warehouse architecture with Machine Learning-based predictive analytics provides a comprehensive Business Intelligence framework that supports both descriptive and predictive insights, enabling more accurate and data-driven marketing strategies in the automotive industry.]]>
