Návrh pravidel a best practices pro tvorbu sémantických modelů ve vybrané společnosti

This thesis addresses the standardization and performance optimization of semantic models built on the Microsoft Tabular architecture, with a focus on the Microsoft Power BI platform. Given the technological compatibility, the proposed procedures are also applicable in the Analysis Services environment. The main objective of this thesis is to define comprehensive rules for designing optimal models, internal structures, and storage mode selection, and subsequently to validate this corporate standard in practice at Seyfor. The proposal is based on an analysis of the VertiPaq in-memory engine and summarizes best practices in the areas of data modeling and the DAX query language. Based on an analysis of the current state of development, supported by semi-structured interviews with experts, a methodological decision tree and a comprehensive catalog of 19 specific rules are proposed. These rules are transformed into a machine-readable format for the Best Practice Analyzer tool, enabling comprehensive quality control. In the implementation section, the proposed solution is applied to a real-world enterprise project in the banking sector that exhibited significant technical debt. The benefits of optimization are precisely quantified using the GQM framework and the Performance Analyzer tool. The results demonstrate that automated error detection and the subsequent correction of structural and computational deficiencies led to a reduction in the total visual loading time. The thesis thus provides a comprehensive, data-driven framework for eliminating technical debt and offers the possibility of future integration into CI/CD processes.