Madlab Day 6/7&8!
Things have moved on a ways since the last time I posted about the placement. After generating some more responses to our sruvey in the last sessions we then got together and took a look at the data. After a few hours over some fries and ginger beer (in my case) in Common we found that once we started trying to write a Python script things went awry. This is because as we had allowed some free text answers in the survey the program we wrote was unable to generate meaningful responses from here. Group names came out as all sorts of rubbish!
The solution to this was unfortunately to start over. This did however allow us to build in more sustainability to the project as we moved over to Google Docs, and a Google Survey as our data collection tool. This is more long term, as we are no longer using Kathleen’s account to gather data. From here we then generated the kind of exel docs we need to feed into Gephi, when we want to create the initial visualisation.
We each contributed to create lists of groups, external agenices and types of connections between the groups. An example could be ‘Girl Geeks’ ‘funded by’ ‘arts council england.’ By creating these lists the survey only uses drop downs, eliminating the need for free form text responses that were causing us so much trouble in the next stage.
This time after having had a few frustrating moments trying to get Phython working last time, we got on a roll. Party due to MadLab connections in action! As we sat in MadLab oppostie a Phython expert… we were able to use his expertese to help us finish the Phython we needed to get our data from Google, into xml and into something Gephi understands!
See the image below for DJ, our hero of the hour!
Between meetings Dave tidyied up the last few issues we had with Python and Kathleen inputted the repsonses we had had from the previous survey into the new one. With Dave tied up in MadLab events this week, Kathleen and I got together to have play with the Gephi graph we now had that contained our actual responses!
So the following screenshots detail the different visualisation modes I tired out in Gephi, as Kathleen tackled some of the parts of the data that were still tripping up the Python.
This provides us with the connected nodes centrally, with the other non connected nodes scattered around it.
Yifan Hu Proportional
Connected nodes int he middle, other scattered in a more spaced out fashion around the outside.
Continuing to run it creates larger spaces between the outer isolated nodes.
This contracts what was already there into a circle. It keeps going until it’s very tight.
Force Atlas 2
This contacts just the mid, edges section.