When reviewing Ph.D. student applications, I look for students who have:
(1) a strong background in cognitive science,
(2) a strong math background (probability, statistics, machine learning, data mining), and
(3) research experience and publications.
I do not accept students who wish to do pure machine learning. I support student focused on 1) educational data mining 2) software engineering that will be used to create tools that can be used to run randomized controlled trials on student learning and 3) folks that have some teaching experience.
Two students of mine that symbolize combining the first two goals are Seth and Xiaolu, where each built their own systems (PLACEments and ARRS respectively), and both have lots of skills that allow them to: 1) do predictive analytics (educational data mining, like how does the data from this feature help us better predict state test scores or student knowledge), and 2) run randomized controlled trials. For PLACEments we have many RCTs going to answer research questions of the following nature: i. Is a particular feature good? ii. Can we figure out what arcs in a prerequisite graph are good and useful in and of themselves? A feature is defined as "good" when it works, improves student learning and teachers see value in it. We also use its popularity when determining goodness. For ARRS results check out this here.
Prospective students should read my recent papers and look at my recent activities (workshop presentations, etc.) to determine if there's a good match of interests.
I respond to email inquiries only if our mutual interests seem strong and the student has the qualifications listed above.