PhD Update

The latest synopsis of my research is this short paper entitled: ‘Ways of Seeing – Student Learning & Metacognition using Machine Learning and Learning Models‘. It’s still relatively early in the PhD process and many theoretical and methodological details remain to be finalised. But some of the values driving this research include:

  • Prioritisation of the student perspective and the student voice
  • The use of data to shine a light into student experience and provide alternative ways of seeing student activity
  • Promotion and capacity building around data literacy for students
  • Application of Machine Learning techniques to help students identify their own learning path

Right now, I am looking at processes and data in a classroom at IT Sligo on our ‘Internet of Things’ module and will be looking at how we can present this data to students so they can make inferences about their own learning process.



Reflecting on LAK16


Arthur’s Seat – looking down on Edinburgh

And there’s a hand, my trusty fere!
And gie’s a hand o’ thine!

Just over a week ago, the first workshops of the Learning Analytics Knowledge (LAK16) conference got underway at the University of Edinburgh. I’ve been getting up to speed with research in this area for the past few months – so it was a great opportunity to get a close-up look at how the field is evolving and how I might be able to situate my PhD contribution in that bigger context. It was also a wonderful opportunity to meet some people in this dynamic and diverse community of learners and researchers – some of whom I’ve known on Twitter for quite a while. This post is an attempt to capture some thoughts from the event and perhaps work through some questions in subsequent posts.

Sheila McNeil has already written an insightful post comparing the #LAK16 event with Open Education Resources #OER16 – another education conference which took place in the same venue the week before. At first glance, these events seem to be quite different: the emphasis on building a culture of openness at #OER16 feels like quite a contrast to the evidence-based, data-centric feel of #LAK16. It could be argued that one takes the philosophical stance of opening up space, where the other attempts to close down ambiguity and open-endedness in the messy business of learning, with its emphasis on data, measurement and rigor. But after a few days of reflection, I think that’s a simplistic view and I wonder what these two fields can learn from each other.

It was interesting that the two opening keynotes, Catherine Cronin at OER16 and Mireille Hildebrandt at LAK16 both underlined the human aspects of learning – the potential for our humanity to be augmented in new learning spaces as well as the threats to our dignity, liberty and equality. A tweet at LAK16 expressed some incredulity that the term ‘human rights’ would be a used in a discussion about Learning Analytics and I have to say I wondered about this. I look forward to further conversations about this observation.

Learning Analytics, or ‘statistical analysis‘ as David Wiley likes to call it, is an emerging field. For a primer in the domain, I recommend Rebecca Ferguson’s paper from 2012 which sets out an informative historical perspective and identifies some of the sub-domains. We as learners and teachers, are producing ever increasing amounts of data – in Learning Management Systems (LMSs), on social media, in institutional systems. Industry and other fields like healthcare and finance have been using these kinds of data ‘exhausts’ as part of a continuous improvement cycle to increase efficiency, improve performance and augment bottom lines for a long time. So why not do the same for learning? But of course, learning is different.

But shouldn’t we be doing something with all this data? This starting point of: ‘we have data – so let’s do some Learning Analytics’ was critically questioned by Abelardo Pardo (University of Sydney) at the outset of LAK16. He encouraged attendees to ask instead what problems we want to solve in education and how can data and analytics help us to solve them? This prompt really resonates with me: how can we bring our critical faculties to bear on the potential for data to do good in our learning spaces? What are the problems we see? What are the problems that we cannot see? One of the affordances of data is that it makes the invisible and the unquestioned visible. What are the problems we can make visible and solve – this is our question, surely?

Ethics and privacy are key concerns in this field and LAK has a history of giving this topic due consideration and thought. This year was no exception and one of the highlights was a paper by Drachsler & Greller which builds on the sterling work of Prinsloo & Slade at previous LAK conferences and provides a checklist to institutions and individual teachers and researchers to help build and maintain trust in the Learning Analytics process. For me, this question deepened in terms of complexity during the conference – I became more aware of the safeguards we need to ensure learners’ interests are protected – but I also appreciated the opportunities that are opening up to empower students to become more active agents in their own learning journey. As learners, teachers, data scientists and researchers, we have a delicate balance to manage. Err too far on either side, and we will waste these opportunities or do harm.

Student voice was in short supply at LAK12 when Audrey Watters wrote about that event four years ago and I have to report that has not changed significantly. One presentation stood out as an exception here: Jen Tan from Nanyang Technological University, Singapore presented work from a second level institution where young learners were picking up some of the skills of networked learning. Their voice came through loud and clear, describing how their classroom visualisation ‘makes me more motivated to comment so that my [social network] dot can be bigger and brighter’. There are many interesting interpretations of what students are telling us here.

Many worry about the role of the teacher in the classroom of the future, but Aneesha Bakharia and her colleagues at Queensland University of Technology (QUT), Australia put the teacher right at the heart of their mixed methods research. They used qualitative research to dig deeper into the experiences of teachers and their learning design practices and how these related to learning analytics toolsets. It’s a paper I’m looking forward to reading in more detail both from a research question point-of-view and from a methodology standpoint.

The inimitable Doug Clow

Lots more I want to write about – maybe in another post… or two:

  • LAK Failathon – talking about failure – we should do more of this
  • Erik Duval – a wonderful tribute at LAK16 to one of its founder and most influential members
  • CLA Toolkit – bringing learner data together from disparate sources and the #xapi data format
  • PELARS Project – wonderful work in Problem Based & Informal Learning spaces
  • The Broad Church that is LAK
  • The need provide data literacy capability-building opportunities for learners and teachers
  • The use of Bayesian Networks in modeling user learning a theme picked up by Robert Mislevy in his keynote
  • That Mark Glynn from DCU and I were the only attendees from Ireland was a surprise

And of course, some of the wonderful LAK people I met in Edinburgh – Aneesha Bakharia, Kirsty Kitto, Sheila McNeill, Doug Clow, Martin Hawksey, Garron Hillaire, Daniel Spikol, Tore Hoel, Dragan Gasevic, Vania Dimitrova, Jeff Grann, Shane Dawson.

… to be continued



Ferguson, R. (2012). Learning analytics: drivers, developments and challenges. International Journal of Technology Enhanced Learning, 4(5/6), 304.

Drachsler, H., & Greller, W. (2016). Privacy and analytics: it’s a DELICATE issue a checklist for trusted learning analytics (pp. 89–98). ACM Press.

Buckingham Shum. (2016). #LAK16 Hildebrandt keynote: A world first? #LearningAnalytics uttered in the same breath as “Human Rights Infringements” [Tweet] Retrieved on 05 May 2016 from

Cronin, C. (2016). ‘Open Culture, Open Education, Open Questions’ Keynote #OER16 retrieved 05 May 2016 from

Hildebrandt, M. (2016). ‘Learning as a machine. Cross-overs between humans and machines’ Keynote #LAK16 retrieved 05 May 2016 from

Watters, A. (2012) ‘Learning Analytics: Lots of Education Data… Now What?’ retrieved 05 May 2016 from

Tan, J. P.-L., Yang, S., Koh, E., & Jonathan, C. (2016). Fostering 21st century literacies through a collaborative critical reading and learning analytics environment: user-perceived benefits and problematics (pp. 430–434). ACM Press.

Bakharia, A., Corrin, L., de Barba, P., Kennedy, G., Gašević, D., Mulder, R., … Lockyer, L. (2016). A conceptual framework linking learning design with learning analytics (pp. 329–338). ACM Press.


Media Literacy to Digital Literacy to Data Literacy

My undergraduate degree was in Communications Studies at Dublin City University (DCU – then known as the National Institute of Higher Education, Dublin NIHED). There I was fortunate to meet lecturers like Luke Gibbons (Film & TV Studies), Martin Croghan (Linguistics), Des Bell (Media Studies) who opened my eyes and my mind to the way in which the media was shaping our world. We got a grounding in subjects like Sociology – where I first encountered Mead and Freud; Economics where I was introduced to Capitalism and Marxism. We had a Psychology class which featured live performances of Woodie Guthrie songs complete with accompaniment from a real fascist-killing-machine. It was quite an education and gave me a jump on media literacy and critical thinking  which has profoundly shaped the way I see and interpret the world ever since.

In the following years, I was lucky to get an equally good grounding in digital literacy and the ways in which networked computing has shaped and continues to shape our world and our culture. Many of these changes are for the better. We are connected in new and rich ways. We can collaborate and work with colleagues and collaborators all over the world. We are more open to each other’s culture and we can appreciate each other’s difference better than we used to. Now, you could argue that the converse of all these statements is true also. And this exemplifies exactly the complexity that is now part of all our lives. If we can build capability around diversity and remain open to contradictions in this new world, we have huge opportunities to evolve. And yes, the converse of that is true too.

Paulo Freire
Paulo Freire

Digital literacy for me was also bound up with learning – and I was influenced by the writings of Stephen Brookfield, who in turn led me to the writings of Paulo Freire. I started reading ‘Pedagogy of the Opressed’ back in 2007, I think. And I read it slowly – savouring every page. In fact, I still hadn’t finished it last year and my promise to myself was that if I ever did finish the book, then I would have to start living it more fully.

It feels like we are at the start of another phase of technical innovation which is fundamentally shaping and molding our lives and how we exist in the world: the age of data. To thrive in the future, we will need to be data literate. We need to understand how data is being leveraged to determine our access to education, employment, finance and so many other aspects of our well-being. The power dynamics don’t seem to change much – but the medium does. It’s up to all of us to get to grips with this new one.



MOOCs in Machine Learning & Data Science

During these first months of my research, I am trying to get a foothold in many technical topics that are relatively new to me – or that I have not worked with for some time. I am working within a Machine Learning research group – Machine Learning being a subset of Artificial Intelligence. Learning Analytics also requires a grounding in Statistics and newer approaches and tools in this area like Python and R. So, there’s a lot to learn, and quickly. I have found some very useful resources in the form of Massive Online & Open Courses (MOOCs) on these topics.

I find it interesting that many of the so-called xMOOC (Siemens 2012) platforms were founded by scholars in the field of Artificial Intelligence and Machine Learning. For example, one of the first high-profile xMOOCs was ‘Introduction to AI’ run by Sebastian Thrun and Peter Norvig from Stanford University – which enrolled 160,000 students from around the world. This was followed by courses in ‘Machine Learning’ by Andrew Ng and ‘Introudction to Databases’ by Jennifer Widom. Out of these early initiatives emerged Udacity founded by Sebastian Thrun and Coursera founded by Daphne Koller & Andrew Ng.

Some of the courses I have looked at or enrolled include:

The way I have engaged with these courses varies – some are an opportunity to refresh on basics. Some are providing me with very practical skills which I am starting to apply to real-world data. Some are just filler content. In general, I have not interacted with other students – although I would normally value this aspect of the learning experience. I am also taking a face-to-face course at my university on Linked Data and the Semantic Web. Again, I am surprised that I actually feel less involved and interactive with other students than I would normally experience in an online learning classroom.  I am wondering is this because of the architecture of the courses? Most of my previous experience of online learning has had a big component of student-to-student interaction. But these courses privilege student-to-teacher interaction. Interesting…


Siemens, George. (2012). “MOOCs are really a platform”. Elearnspace. Retrieved 2015-Mar-15


My Research – Learning Analytics, Data Mining & Machine Learning

Dr. Michael Madden

In October 2015, I was delighted to start a PhD with the Machine Learning & Data Mining research group at the National University of Ireland, Galway. The group sits in the IT Department and my supervisor is Dr. Michael Madden. My research interests lie at the intersection of learning and computer systems – and in particular – Learning Analytics.

Over the past 10 years we have seen the widespread deployment of Learning Management Systems (LMS) like Blackboard & Moodle across our education sectors. We are also seeing learning reach outside the individual institution with the evolution of learning on social media platforms and Massive Open Online Courses (MOOCs).

With Hamda Binte Ajmal & Ankita Garg – colleagues at the Machine Learning & Data Mining research group at NUI Galway

While we could argue about how effective the adoption of these systems has been, there is no doubt that they are changing how we learn and how we teach. They are generating vast quantities data and revealing new insights on how learning happens. We are seeing a powerful force emerging – but with great power comes great responsibility. The choices we make now in how we harness this data will have a profound effect on the way we see learning, the way we think about it and the way it will shape our lives in the future.

Learning spaces comprise threads of developmental connection between learners and teachers. From a teaching point-of-view, they require what Stephen Brookfield referred to as the ‘Three Rs’ of teaching: Respect, Responsiveness & Research. When we make interventions there, based on purely quantitative measures, there is a danger that we could damage these threads. We also face profound ethical issues around privacy and freedom when we consider effective use of learner and teacher data.

Despite these concerns, I am optimistic about the potential of Learning Analytics for good – in our schools, universities, CoderDojos and learning spaces. In this blog I hope to document my research into this potential.

Thank you for reading.