Abstract

National Statistical Offices (NSOs) are increasingly challenged to demonstrate the value of official statistics within a rapidly evolving data ecosystem shaped by artificial intelligence (AI), data science and emerging analytical technologies. Although substantial investments are made in the collection, processing and dissemination of official statistics, opportunities remain to unlock additional value from existing datasets through innovative reuse, interdisciplinary collaboration and the development of new data products. Datathons offer a promising avenue for National Statistical Systems (NSSs) to move beyond conventional dissemination towards a more collaborative, innovative and impact-oriented approach to statistical data use.

Datathons provide structured environments in which statisticians, data scientists, researchers, software developers, students and policymakers collaborate to explore statistical datasets and address real-world challenges. By applying statistical methods, machine learning, geospatial analytics and data visualization techniques, participants can identify patterns, develop predictive models, generate new analytical insights and prototype decision-support tools that complement established official statistical outputs.

The proposal is development a framework for integrating datathons into the statistical value chain, from identifying priority policy questions and preparing accessible, well-documented datasets to facilitating multidisciplinary problem-solving, evaluating outputs and supporting the adoption of promising innovations. Drawing on the institutional context of official statistics in Kenya, the proposal is to consider how NSO-led datathons can promote the innovative reuse of official datasets while strengthening collaboration between data producers and users. It highlights the potential of this approach to expand the analytical applications of official statistics, strengthen data science capabilities, enhance engagement with emerging data users and support evidence-informed policymaking.

This approach contributes to international discussions on the future of official statistics by positioning datathons as a complementary mechanism for fostering innovation within National Statistical Systems. It argues that the value of official statistics should be assessed not only by their production and dissemination but also by their capacity to enable discovery, generate new knowledge and inform decisions. Institutionalising responsible data reuse and collaborative innovation can help statistical offices maximise the public value of their data assets in the era of AI and data science.