Q&A not answered during the session
Q: Are there any differences in terms of data points used as "red flags" between courses or majors? In other words, are the same data points used to identify at risk students in a biology course compared to an english course? And also, are there some courses where the data is insufficient to identify at risk students?
[Sandeep] : That is a good question. There is a generic model which uses standard set of datapoints which are widely collected across schools/majors. If we were to deploy a single magic model :) across all courses & majors then we will have to be restrictive about the data point selection process for predictions to get good accuracy.
We also have been developing a model library catering to various cohorts, preferably on a program/school basis. By taking this approach we will be able to leverage additional datapoints which are specific to school/program. Going back to Josh's example in the webinar, the school of management at Marist are heavy online users. This enabled us to customize/tune the generic model even further to boost accuracy by bring in supplemental datapoints like discussion forum participation etc. .
Q: Can the LRS and LA Processor examine step-level data in LTI-based, third-party apps?
[Gary] Generally speaking if the application can make the activity data available in a supported format it can be processed and stored in OpenLRS and/or the Learning Analytics Processor. We would need more details to provide specifics but it is a fairly typically use case.
Q: Is there a student interface that allows self-directed response to the data?
[Gary] OpenDashboard does support student access (data access is restricted to only the student's own information) but we don't as yet have support for a self-directed response. We'd love to get your requirements and work with you on implementation.
[Sandeep]: Adding to Gary's thoughts - In terms of dashboard we generally think of perspectives (Student/Faculty/Academic Advisor/Management). The initial focus within Apereo LAI was to have the OpenDashboard integration with the rest of the systems in the Diamond architecture (as shown in Gary's demo which was on a faculty level). The next logical step would be to map out and build targeted use cases for each of these perspectives. It would definitely make a great collaboration opportunity within the community to build on the existing framework.
[Alan]: Adding to Sandeep's thoughts - Hackathons have the potential to play an integral part in requirements gathering. For example there was recently a hackathon at the University of Amsterdam that spent two days reviewing SMART goal setting. At LAK15 we there will be an exploration of the core requirements of a dashboard vs specific widgets.