Mapping ridership on a linear corridor
I saw something on the internet a couple days ago:
…that made me want to analyse ridership stop by stop, and well, hopefully make more sense of it than this chart did. It’s kind of pretty, and I’d probably get more out of it if I were familiar with the system, but it’s hard to see any interesting trends here, mostly because the vertical scales vary quite a bit1.
Being the techy that I am, I wrote up a little PHP script to manipulate SORTA’s not-quite-public ridership data from 20092. Stops are paired up, as well as I could manage for each direction. The green bars are passengers getting on, the blue getting off. Grey bars between them indicate the day’s total. Each side of the center line is a direction of travel.
I made a few of these graphs and they got me thinking about what I should have expected the data to look like. I’d never really thought about what transit ridership should look like theoretically. Did you notice that the balance of riders getting on or off shifts completely from one end of the line to the other? It makes sense. If a transit line were to travel straight through an area with even density, we might expect it to look something like this:
We see a shift in boardings and de-boardings because a line is more useful if it has a long way to go yet, and not very useful at all if you’re at the second to last stop. Not many people will get on at 7th St to go two blocks to Government Square but if they get on at 7th to go the other direction they have the whole world ahead of them.
And so we see each trip slowly shift from “on”-heavy at the beginning to “off”-heavy at the end with a solid mix of boardings and de-boardings in the middle.
Here’s the #33 again, in map form, and only going out from Downtown:
Here it is zoomed in on the Glenway/Warsaw business district:
Here’s what just that looked like in the chart by the way:
It’s only the outbound trip, so it’s only one side of the graph. If you look at an aerial photo, you can see pretty quickly that the ridership is higher where there are businesses and dense side-streets going off perpendicularly to houses. The places where riders drop off rather quickly are places where hills have precluded development and one or both sides of the street have either a lot less development within a couple blocks, or even just woods. The hill up from Lower Price hill(East side of the image) demonstrates this effect most notably.
Dense areas are good places to run transit. It’s also good to connect dense areas that are divided by a sparse area like Queensgate. The stops along the way won’t show it, but the huge number of people who loaded on at Government square are still on the bus heading west to the Warsaw Glenway business district and beyond.
There’s a ton more to discover in here but it’s actually a lot of work preparing the data. If you want to do a little exploring yourself, here’s the script and the data that I used. The full stop-level ridership dataset from March 2008 is available on the Data page of this site and a more up to date version will be available very soon if enough people email Kevin, SORTA’s senior system planner. If you haven’t seen it yet, you may also want to check out the full ridership intensity map I made from the same 2008 data. Here are a couple more charts just to whet your appetite:
And the #51:
It’s important to note for these two that it’s not that they’re necessarily “performing” a lot worse than the #33, rather their frequency is significantly lower so less riders could have been expected from the beginning.