Researchers’ Blog Immigration Pipeline, mock up v2

[This builds on previous posts Immigration Pipeline, Dataset for the Pipeline Visualization, and Data Formatting for Pipeline Visualization]

I guess I shouldn't be too surprised that what I thought would be a nice simple depiction of the data becomes infinitely more complex and messy when applying design ideas to the actual data. Today I spent some time playing around with the 5 districts and leading Canadian cities Ollie identified in his data work last week.

This first one shows the 9 Canadian cities Ollie identified, grouped by province. I added the middle layer of registration locations (just that were applicable to these combinations of origins and destinations) -- thinking that it would help organize and consolidate lots of branches coming out of the origins and going to the destinations.

This could be cleaned up a lot more -- i.e. angle and bunch the "pipes" together instead of having each connection be a unique path. But even then I think this is way too busy, begging the question "How can we simplify this?"

My first inclination was to reduce the number of destinations in Canada. I took out the western ones to focus on the remaining eastern ones from this list.



So in this second version there are just the 5 origins (birth districts) and 4 Ontario cities and Montreal. I also consolidated the branches so that there are shared "pipes." Hopefully the slightly transparent colors from each origin helps illustrate this (not intended to be in the final version).

The lines are based on the actual paths between each origin and destination. So for instance, there aren't any immigrants from Hoi Ping to Windsor -- that is why no line connects them directly. But there are (many) immigrants from Sen Ning going to Montreal -- so there is a direct line there. Adjusting width of these lines could further illustrate which paths have the most "traffic."

The other addition to this version is the proportional symbols for the origins and destinations based on the number of people at each place. The bottom row of pie charts are divided by the number of immigrants from each origin. Ideally both sets would change dynamically in correspondence to the animated timeline (e.g. origin circles shrink as time goes by and destination circles expand -- and pie charts grow).

Another element I'd like to consider more is how the timeline is positioned -- and if there are any other elements like a bar chart of immigration totals for all of Canada we want to line up above it. I got stuck with the 975 px wide format because of trying to fit the 9 destinations on earlier, but now I'm wondering if this second version should be collapsed and stretched vertically a bit so that there could be a vertical timeline instead of a horizontal one. One of the things Erik and I were talking about yesterday with this is including other types of information like occupation. Having extra space on the side (either right or left) might be a good place to put a dynamic divided bar chart -- like in Jon's butterfly extinction visualization: http://www.stanford.edu/group/spatialhistory/cgi-bin/site/viz.php?id=19&project_id=0.

I think this is getting closer to the original subway map idea but I'm still concerned that the little dots going from origin to destination may look really busy and not really effectively illustrate any interesting patterns -- "chainness" or otherwise. I guess there's only one way to find out -- start playing around with it in Flash. But in the meantime, please throw around other ideas. I'm not wedded to this concept but am feeling anxious about making sure we nail our December deadline!

These versions are different layers of the same Illustrator file saved here: \\Sul-koala\spatialhistory\Lab_Projects\Collaborations\Henry Yu\Storyboards. Feel free to make a copy and modify it with your ideas.

Immigration Pipeline v1

Immigration Pipeline v2