Analytics Tools

There are two large collaborative projects in the LMS world that hold great promise for collecting more robust learning data that can help power learning analytics.

1. Tin Can API  by ADL (the creators of SCORM) & Rustici Software

Description
Tin Can captures learning experiences in RDF triples (with domain-specific vocabularies) that constitute independent learning record stores (LRSs) that can be read and written to by any number of applications.

Opportunities in Learning Analytic Solutions
LRSs can share data independently with learning apps and tools for reporting and analytics.

Weaknesses/Concerns/ Comments
Creating broadly-accepted domain-specific ontologies and getting buy-in from developers and practitioners are enormous challenges.  

2. Caliper by IMS Global (the creators of the LTI standard)

Description
Very similiar to Tin Can, but is still not in active production.

Opportunities in Learning Analytic Solutions
Promises measurement of learning activity and its performance at a very granular level.

Weaknesses/Concerns/ Comments
The concept is strong, but there has been little published progress in the past 12 months.

Visualizations in Tableau: Part 2

So, Harvard and MIT have released the de-identified learning data for their first-year courses on the edX platform.

I grabbed the data set and made this visualization that lets you see the completion rates by country and filter on level of education and gender (click on the horizontal bars).

I saved it to Tableau Public to get the online interactivity. Click the image below to open the interactive visualization.

Visualizations in Tableau: Part 1

As a means of getting comfortable with Tableau, I connected to the MySQL database for the Russnet online language learning modules. The following packed bubbles visualization shows the top performers on Russnet by first name. The size of the bubbles correlate with the number of correct answers each learner has produced in the online activities. The range on the answer data spans from 3,500 to ~36,000 correct answers.

Introduction

I am an applied linguist in the area of second language acquisition (SLA) with a background in computational linguistics and programming.   My work focuses on how people learn languages and the extent to which natural language processing tools and techniques can facilitate more effective learning.   For the past ten years, I have been drawn to the notion that open, independent learner models can empower rich data visualization, effective formative assessment and individualized learning pathways.

This blog has initially been setup to broadcast ideas and discoveries on learning analytics for the edX course:

Data, Analytics, and Learning