March 15, 2013

Coursera Data Analysis MOOC: Graduation

I've completed Coursera's 2013 Data Analysis course. You can read my earlier posts on the course here and here.

I was initially motivated to enrol so I could learn about Massive Open Online Courses (MOOCs). Once the course started I realised it would take significant effort on my part to see it through. I could easily have given up but decided to invest the time needed to complete the course.

I'm glad I did because I gained the following:
  • knowledge of how Coursera works
  • a broad overview of the statistical techniques that can be used for data analysis
  • improved ability to use R - a tool I often use for work
How Coursera Works
I was out with friends one evening mid-way through the course, and mentioned I'd enrolled with Coursera. I said that the course was free, and was asked "What is Coursera's business model?" I didn't know at the time but I've since read that various revenue streams are being considered:
  • certification fees
  • introducing students to employers and recruiters
  • tutoring
  • sponsorship
  • tuition fees
According to Wikipedia, Coursera was not generating revenue as of March 2012.

The mechanics of a Coursera course are similar to those of a college or university course with the difference being that it takes place online and thousands of students are enrolled.
  • lectures: content is presented as video lectures.
  • quizzes: regular online, multiple-choice tests must be completed.
  • assignments: assignments are submitted online. The lecturer can't assess them all, so students mark their peers' work.
  • getting help: the lecturer can't answer all questions so students post queries to an online forum. Students help each other out with answers, and each course has a handful of knowledgeable TAs who monitor the forum and post replies. You can vote up a post - those with the most votes are handled with the highest priority.
  • wiki: each course has a wiki to which useful course-related information can be added.
  • meet ups: if you want to take things off-line, MeetUps can be organised to discuss the course face-to-face with fellow Courserians
Data Analysis Course Content
Ultimately, the quality of a course, whether traditional or MOOC, hinges on its content. A friend of mine, who is a university maths lecturer, enrolled in a Coursera programming language course but found the content so poor he gave up.

Overall, the Data Analysis course was good quality. It was the first time Prof. Leek had given the course so there were a few mistakes in the course material. These were picked up by students, who posted corrections to the online forum.

There were also logistical difficulties for students in some time zones. To accommodate them deadlines for quizzes and assignments were tweaked.

I expect the course will be given again, so future enrolees will enjoy the benefits of the road-testing performed by my cohort of students.

Data analysis is a very broad subject, so it was difficult for Prof. Leek to provide a detailed presentation of the techniques covered in the eight-week course. Instead, a basic introduction was presented for each technique, with examples of how to perform the analysis using R. Links to further resources were provided for those students with the time and inclination to delve deeper into the underlying mathematics. This was something I didn't have time for but at least I now know where to start.

Coursera offers a broad range of courses, and then there are courses offered by others. Having completed my first Coursera MOOC I'm tempted to enrol in another but they do require a significant investment of time and effort. For now I'm content to consolidate what I've learned and wait for something to come along that piques my interest sufficiently for me to put in the effort required.