Creating a Language Placement Dashboard

5 December 2017
By Solomon Foster

This term I have been working with some academic technologists to start the process of creating a dashboard for language placement exams. This most recent meeting is a continuation of the work Quinn talked about in his post concerning sketching a backend for language placement tests. Some challenges and considerations on the technical side are as follows:

  • Best way of visualizing agglomerated test data: We decided on heatmaps. Heatmaps are versatile in that rows and columns can represent a variety of levels of different variables, from very granular to more high-level. For instance, in the case of the language dashboard, each row in the heatmap might represent an individual student, a cohort of students, or an entire class. Columns might represent individual questions or question subject-areas (i.e. vocabulary or grammar) and might be ordered by question number or difficulty.
  • How to allow the dashboard to interface with Moodle: As far as I can understand, Moodle doesn’t really have a client-facing API. It is accessible through PHP and some other wizardry, but as a group we decided this dashboard would not try to interface directly with Moodle. Instead, the academic technologists from Swarthmore, Haverford and Bryn Mawr advocated a data dump (or cron job) every so often into a local csv or excel file, which would then be read in by the dashboard (easy to do with D3).
  • Hosting the dashboard: Some debate about whether the source code for the dashboard should be sent to individual professors or managed by some central authority — or blockchain!. The consensus was to have the dashboard managed by an academic technologist somewhere with knowledge of D3, and then potentially shipped out to professors in one directory. Could we create an executable? Probably not. Executables sometimes break for no reason.

Overall, an interesting experience working with professionals running the gamut from qualitative to quantitative. The least quantitative people I’ve developed anything for knew SQL already, so I imagine the design process with language professors will be iterative.

DataSquad Students Advising at LACOL
DataSquad members Oscar Smith ’20 and Solomon Foster ’18 advise at a LACOL brainstorm.