Engineering and R&D

Learning analytics for an online course

A full CRISP-DM study of seven runs of a FutureLearn MOOC, from 61 raw files to a clear picture of who signs up and who stays.

Year
2023
Context
Data science, Newcastle University
Tools
R, ProjectTemplate, dplyr, ggplot2, R Markdown
Pie chart of the ten most common employment areas among learners, led by IT and information services.

The aim

Use CRISP-DM to understand the learners on a massive open online course (MOOC) run on FutureLearn, across seven runs of the course.

What we did

  • Triaged 61 raw data files down to the 7 that answered the business question, then documented why each one was kept.
  • Built a reproducible analysis in R with ProjectTemplate, dplyr and ggplot2. The report is written in R Markdown, so it rebuilds from the raw data in one step.
  • Followed each CRISP-DM stage in turn, from business understanding through data preparation to evaluation.

What we found

  • IT and information services was the most common employment area among learners, followed by teaching and education, then the public sector.
  • The share of employed learners fell from run to run, except in run 5.

These patterns tell the course team who to design for and where to promote the course.