About this Dataroom
For every pair of members on the Toronto City Council, three values are given (twice in the file since they are symmetric):
- The number of items voted on where their votes coincided
- The number of items voted on where their votes differed
- The total number of items both voted on (sum of the previous two)
The source is from an aggregation of voting records found here:
- http://buzzdata.com/haz/toronto-city-council-voting-records
The voter correspondence is computed using a hierarchical clustering algorithm with a ward linkage type (0). The clustering result can be found here:
- http://www.haz.ca/images/viz/linkage-ward.png
...and the projection to the map of Toronto can be found here (note the colouring for both is the same):
- http://www.haz.ca/images/viz/toronto-wards.png
postbear
momoko
Note: This update might be based on outdated information.
The voting data is likely out of date by now -- until there's a way to export all of the records on mass, updating will be rare. I had to select / download the csv for each councilor individually.
Thanks for the expansion! Any visualization in the works?
Did you sort the rows based on difference? (after sorting based on the first column name)
I sorted first based on city councillor, then based on similarity (descending). Was happy to note that my councillor (Mike Layton) is most different from Rob Ford :)
Ahh, right on. It makes it easier to see things that way.
If you sort based on the % difference, you can see the most extreme sides being Rob / Doug Ford on one side, and Paula Fletcher / Anthony Perruzza on the other (8 / 8 of the top differences are between one in the first group and one in the second).
Would you be interested in uploading my expanded version of your dataset? (Until we can do merges, I'll have to resort to this!) Link above ^
sarah
crimeminister
Great dataset and tree visualization, which reminds me of tournament fight card. It's only missing some pugilistic profile photos with angry glares and furrowed brows.
It might be cool to label the tree edges with the "Agenda Item Title" from your voting records dataset that occurs the largest number of times in the set of votes taken by a given pair of councillors where they voted differently. That might provide a useful hint as to the topics on which any two councillors are differing the most.
Hrmz. So the most contentious issue for any grouping of candidates (the opposite definitely wouldn't work, since there are items that every councilor works on). There may be a problem since some of the contentious items may live in the sub-councils they sit on (which only sees mutual voting with a handful of other members).
Is not!