Abstract

Enterprise business intelligence relies on dimensional modeling because star schemas provide a business-friendly semantic layer, consistent metric definitions, and efficient query performance (Kimball & Ross, 2013). As finance, sales, and operations require shared measures such as revenue and active customers, direct querying of operational schemas can produce inconsistent table joins and business rules. Dimensional modeling addresses this by defining clear grain, organizing measures in fact tables, and using conformed dimensions to standardize entities across domains (Chaudhuri & Dayal, 1997). Dimensional models are derived from normalized relational sources through subsetting, denormalization, and aggregation, but students often struggle to convert normalized schemas into analytics-ready star schemas (Agboola et al., 2023). To support this transition, the module uses a structured, prompt-engineering-supported workflow with the TPC-H benchmark dataset to guide students through grain declaration, fact and measure identification, denormalized dimensions, surrogate keys, and ETL in MySQL and BI tools such as Tableau (O’Neil et al., 2007; Wei et al., 2022; Zhou et al., 2022; Kimball & Ross, 2013). By integrating dimensional modeling, ETL practice, and AI literacy, the framework connects data warehouse theory with applied analytics engineering and critical evaluation of AI-supported workflows (Long & Magerko, 2020).

Download

Share

COinS