Sunday 15 December 2013

My first Coursera experience

"Congratulations! You have passed the course “Social Network Analysis" and received a Statement of Accomplishment."…. so ends my first, though most probably not last, experience of attending a course delivered via Coursera.

Given my role in IBM, personal interest in trying new technologies, and enthusiasm for learning new things, it was only a matter of time before I enrolled on one of the emerging group of virtual courses.

Coursera seemed a good place to start and so some months back I browsed the catalog of upcoming courses, found an interesting one that would give me an introduction to the analysis of social networks, and signed up.

Delivered by Lada Adamic from the University of Michigan the course was 8 weeks long.  Each week we watched a series of videos that Lada had recorded where she would take us through some lecture materials and then set us homework to submit and complete.   Homework often revolved around using a modelling tool (such as Gephi or NetLogo) to explore models that she had prepared and then answering multiple choice question based on what network behaviours the model allowed us to observe.  Total time commitment probably averaged out around 2.5 hours a week.

There is clearly a limit to the amount that you can cover in 8 short sessions but I please to report that I very definitely emerge from this experience with a better understanding of Social Networks than I had at the start.   Only problem is that now I want to go and learn more :-) We got a fascinating glimpse into how some of the network modelling approaches could be applied to things as diverse as drug development, recipies, and industry cartels, as well as the more expected Facebook style social networks.

One of the early pieces of homework was to analyse our own Facebook networks so here is a view of what mine looks like.  The vertices/dots are people and the lines show links.  The people included in the graphic are my Facebook friends, the links shown are the links they have to each other. I am not included as one of the dots - this is about the links my contacts have amongst themselves independent of me.  Displayed this way you can start to see how my network has a variety of clusters in it.  As you might expect the other IBMers that I am linked to have a tendency to know each other as well.  Likewise my family members tend to similarly link to each other but much less likely to connect back into the IBM part of my network.