Development of a performance dashboard data model
Saarinen, Juhani (2016)
2016
Tietojohtamisen koulutusohjelma
Talouden ja rakentamisen tiedekunta - Faculty of Business and Built Environment
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2016-03-09
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tty-201602193543
https://urn.fi/URN:NBN:fi:tty-201602193543
Tiivistelmä
Many organizations have noticed the benefit of refining data into information by business intelligence processes and collect a large amount of data from their operative systems. A performance dashboard is used for visualizing data in order to measure the performance of certain processes. A data model is needed between source systems and data visualization to join data from different sources and process data according to the business requirements of the dashboard. Especially the large size of source data may cause challenges for the performance of the dashboard. Thus, decisions related to data granularity and aggregation need to be done. Database design is a method for transferring the business requirements into a technical form.
This research was executed as design science research in the case company. The research consists of a literature review and an empirical part. The literature part consists of theories regarding database design, data warehouse and QlikView architecture, QlikView data model, performance testing and data visualization. The data for the empirical part of the research was collected through participant observation during the project. The theoretical part indicated methods and data models that can be utilized in data model design. In addition, best practices in QlikView performance optimization, visualization and testing were identified. These theories were applied in practice in the empirical part of the research.
The results of the research included functional and non-functional requirements for the data model, a conceptual model, logical planning for the data warehouse tables and QlikView data model, data visualization in demo dashboard and performance testing results with sample data that fulfilled the performance requirements of the dashboard. The development of the dashboard data model was done by identifying the most important entities and attributes in the data sources and developing a logic for joining the data. The data was aggregated in the data warehouse side due to the large size of source data. As a result,
detailed level ad-hoc queries can be done in data warehouse target tables whereas aggregated data is presented in data warehouse views that are utilized in the dashboard data model. This research demonstrated the concept for the data model and the solution could be scaled into production use.
This research was executed as design science research in the case company. The research consists of a literature review and an empirical part. The literature part consists of theories regarding database design, data warehouse and QlikView architecture, QlikView data model, performance testing and data visualization. The data for the empirical part of the research was collected through participant observation during the project. The theoretical part indicated methods and data models that can be utilized in data model design. In addition, best practices in QlikView performance optimization, visualization and testing were identified. These theories were applied in practice in the empirical part of the research.
The results of the research included functional and non-functional requirements for the data model, a conceptual model, logical planning for the data warehouse tables and QlikView data model, data visualization in demo dashboard and performance testing results with sample data that fulfilled the performance requirements of the dashboard. The development of the dashboard data model was done by identifying the most important entities and attributes in the data sources and developing a logic for joining the data. The data was aggregated in the data warehouse side due to the large size of source data. As a result,
detailed level ad-hoc queries can be done in data warehouse target tables whereas aggregated data is presented in data warehouse views that are utilized in the dashboard data model. This research demonstrated the concept for the data model and the solution could be scaled into production use.