Developing an open architecture for learning analytics

Carrying out learning analytics can involve a complex range of data sources, systems and dashboards, often owned by different parts of the institution. Vendors are quickly moving to try to capture the market and ensure their products are at the centre of the learning analytics landscape (Sclater, 2014b). However the architectures for connecting the various elements are still evolving as new applications for learning analytics emerge. Complicating the picture further, the wide variety of systems in place at different institutions means that architectures are not easily replicable between organisations or sometimes even between departments. The interest in deploying learning analytics services at the campus level is increasing but there are many barriers to deployment such as: breaking down data silos, understanding the predictive models and interventions, data governance, and the lack of competences within an organisation needed to manage interventions. There is therefore a growing need for guidance to institutions which wish to develop their learning analytics capabilities as to how to integrate multiple data sources and the new systems they need to build, procure or co-develop as part of a wider community. In the UK Jisc is spearheading an initiative to procure the elements of a basic learning analytics system for higher and further education institutions. This has involved developing an architecture comprising a number of discrete data sources and systems. The model was reviewed by a cross disciplinary team of European experts in Paris in February 2015. This paper describes the Jisc learning analytics architecture and proposes it as a reference model which organisations can use to help develop their own architectures for learning analytics. The experiences acquired and lessons learned during development have the potential to influence and be influenced by wider international discussions around an emerging Open Learning Analytics Framework such as the Apereo Learning Analytics Initiative. The paper demonstrates how an architectural walkthrough with invited experts taking on discreet roles was used to enhance the architectural model.