Enhancing employee data integration in a data warehouse

Abstract

As organizations increasingly rely on data-driven decision-making, ensuring high data quality in enterprise data warehouses becomes critical. This thesis examines data quality issues in the employee dimension in Íslandsbanki's Modern Data Warehouse, Trölli. Íslandsbanki, a financial institution in Iceland, launched Trölli as part of its digital transformation to replace the legacy system Valhöll, which had become unsustainable due to ad hoc development and limited scalability. The work identifies critical shortcomings in the inherited employee dimension, including incomplete coverage of foreign employees, limited attribute richness, and inadequate modeling of the manager hierarchy. To address these issues, the thesis proposes a redesigned schema that integrates data from H3, an HR system in wide use in Iceland, ensuring completeness and improving analytical capabilities. A central insight of this work is the importance of the data exploration phase in data warehouse development. Early decisions about source systems and schema design have long-term implications for completeness, expressiveness, and sustainability. The work proposes a strategy to define a natural key, when limitations of the source system mean that no single attribute has the required qualities to be unique, persistent and mandatory. The new solution allows for accurate historical reporting of all employees and introduces attributes to improve lifecycle analysis. A recursive accumulating snapshot fact table is implemented to represent multi-parent, variable-depth hierarchies. Together, these enhancements provide a more robust foundation for HR analytics and organizational reporting, aligning the DW with best practices and future analytical needs.