Friday, March 09, 2012

I need to get there by 9ish: Refined Toronto transit time map for the TTC and GO

So I've now refined my program for visualizing the transit times for Toronto. First of all, I've now included GO Transit data in the simulation. As I suspected, it didn't have much of an effect on the results. The distances in Toronto are just to small and GO Transit service is simply too infrequent for it to show up in a simulation of this type. Although taking a GO train might get you downtown faster than riding the bus, because the GO train comes so infrequently, you will likely arrive too early at your destination and end up waiting around. Since this simulation tries to calculate the latest you can leave from a location and still arrive at your destination by 9, the GO train ends up faring poorly. It may be possible that my 100m walk restriction may have resulted in GO train stations being unreachable (since GO Transit is mostly designed for car drivers, most of the stations are surrounded by giant parking lots, and you may have to walk more than 100m from the nearest bus stop to get to the train platform), but I don't really see any evidence of this sort of behaviour on the map, and I think the TTC does try to send buses as close to the platforms as possible. I'm not sure how frequent GO Transit service would have to run before it would show an effect on the map. I'll perhaps have to run a simulation one day to figure that out.

Data, imagery and map information provided by MapQuest,
Open Street Map and contributors, CC-BY-SA.

Another change I made was to discard bus stops from the map that do not have any service between 7am and 9am. These are likely late-night bus stops that don't normally receive service during the day, so it was misleading to include them in the final map of rush hour transit times.

And finally, I no longer simply calculate the best way to get to downtown by 9. I also calculate the best way to get to downtown by 8:55, and I then average the two results. This helps ensure that the data isn't skewed because the schedule shows a perfect transfer of buses (e.g. a bus that only runs every 30 minutes, but just happens to arrive just before another rare bus so that you can get to downtown without waiting--even though in practice, if the timing is just a little bit off, things don't line up and you end up 30 minutes late). By averaging the results for different times, it reduces the magnitude of this effect.

In the end though, the final map visualization looks pretty much the same as the previous one I generated except for the extra GO Transit routes.

1 comment: