Quantified Self is a really cool movement of people doing self tracking using technology — for example, one might use a device to monitor their heart data or when they’re at home, and then analyze it. One idea several people around the lab have been toying with is applying these ideas to organizations one is a part of. Just like individuals can benefit from Quantified Self by gaining objective information about themselves, organizations may be able to similarly benefit. (We admit, our motivations mostly boil down to: data  is cool and graphs are pretty.) The natural place to begin, of course, was with hacklab!

We (Sen and Chris) were really excited about this and have done some initial analysis. Hacklab (like, we think, most hackerspaces) had a lot of sources of data laying around, waiting to be analyzed:

• doorbot (in my opinion, our gold mine)
• Google Calendar
• IRC Traffic
• Twitter Traffic
• Mailing List Traffic
• Blog Hits
• Hacklab Public Computer Activity?

So far, we have only worked with the doorbot data.

Activity Levels

Unless the door has been unlocked, entering Hacklab requires one to use a small fob, unique to each member. The program responsible for processing these, doorbot, will unlock the door if it detects a member. It also logs the entry in a database. This is a valuable source of data about activity at the lab, but there are a number of ways in which it can be flawed. If a member works on a project on the side walk outside, they may enter and leave a number of times in a matter of minutes, but this doesn’t actually mean there was more activity. On the other hand, a member may enter along with another or on when the door is unlocked, making them invisible. Furthermore, Fob’s may be reassigned over time, and we have no way to know who the former owner was. The first concern is mitigated in the following data by considering only the number of entries by unique members each day.

Since the data is really noisy, let’s take a 7 day average to smooth it out a bit:

Looking at this, you may be inclined to suspect seasonal trends, but let’s overlap the years to draw them out.

Again, this is really noisy. Let’s do a 5 day sum to smooth it out.

Interesting observations:

• consistent slump in activity at the end of December (shopping/celebrations keep people away?)
• consistent bump in November
• 2009/2010 have slumps around the Summer but 2011 doesn’t

Another thing to look at is the derivative of Hacklab activity:

Gender Distribution

Another subject of interest is Hacklab’s gender distribution.

To avoid making oppresive assumptions, members were given the opportunity to self-identify their gender. We ended up with the gender categories male, female and other. There were very few responses, so most of this data is based on our understanding of how members identify.

The line for male is blue, the line for female red, and the line for other green. Please ignore any sex-stereotyping in this color scheme.

Again, the data is very rough. Let’s smooth it out with a 5 day sum.

I think that percents may be more useful as a mode of analysis.

Hacklab seems to have been getting better in terms of gender diversity lately, though not as good as we were in mid 2009.

More Graphs!

Some more graphs relating to gender and year comparisons are available. Many more will be coming as we move on with Quantified Hacklab! Also prettier ones, since this was fairly quickly hacked together!