Datakvalitetsutmaningar i offentlig sektor : En utvärdering av datakvalitet

Sammanfattning

This paper explores the role of data quality in the development and implementation of predictive models, with a particular focus on applications in the public sector. As artificial intelligence (AI) has increasingly become central to the advancement of predictive analytics, this study also considers AI-driven approaches as part of the broader context. Drawing on prior research and theoretical frameworks concerning data readiness and data quality in big data contexts, the study investigates how organizations manage, prepare, and utilize data for strategic decision-making. The research builds on insights from predictive analytics across various sectors, including education, healthcare, and business, where forecasting models have long been used for planning and resource allocation. By comparing established applications with a case study from the education sector, this paper highlights both shared challenges and domain-specific issues related to data quality. The empirical material consists of qualitative interviews with professionals who either analyze data to support strategic goals or handle the collection, cleaning, and storage of data. As a practical case, the study investigates predictive forcasting aimed at identifying students at risk of transferring between educational institutions. This case illustrates the potential of predictive analytics to support proactive planning in schools, but also the limitations that arise when input data is incomplete, outdated, or inconsistent. The findings reveal recurring challenges such as fragmented data governance, inconsistent quality assurance practices, and a lack of standardized frameworks. The study underlines the importance of integrating domain knowledge, ethical considerations, and interdisciplinary collaboration to enhance the reliability and value of predictive models in complex organizational settings. The paper concludes by calling for more structured approaches to evaluating and improving data quality prior to the implementation of predictive systems. It also suggests that better collaboration between data scientists and domain experts may mitigate some of the risks associated with poor data quality and help create more robust, actionable predictions in the public sector.