I’ve got a long commute into work on a daily basis. That commute gives me a lot of time to sit alone in my car and contemplate the fact that I have a long commute.
In the middle of the night I can drive it in 45 minutes, but with traffic it’s never under an hour, often 1.5 or more. I’ve noticed that very small variations in when I leave, even just 10 or 15 minutes one way or another, can really make a huge impact. But it’s not a sure thing. Some days leaving at 7:45 is the best, some days it’s 8:00. I don’t know why it differs.
It’s my belief there are certain patterns that might take into account things like weather, day of the week, departure time, school vacations, and a bunch of other variables. But I don’t have a good way to that figure that out since I’m lacking the data.
Which brings me to my latest mini-project. I’m writing some mobile software for my phone, plus a website that will help to record my daily commute and visualize what’s happening. Eventually, I hope to bring in some statisitical models to help predict my average commute time based on when I leave the house.
- Phase 1: Start collecting data on my phone
- Phase 2: Collect the data on a website.
- Phase 3: Visualize the data.
- Phase 4: Make predictions based on a single user’s data.
- Phase 5: Collate multiple users who commute in the same area to improve #3 and #4
It’s my hope that someday I’ll be able to answer questions like:
- Tomorrow, what reasonable time should I get out of the house for the shortest commute?
- I’m running late, should I work from home for an hour or head straight in to avoid some traffic?
- What week is the worst traffic-wise of the year? (And hence, what week should I favor for planning a vacation)
- What’s my fastest commute time? Did I break that record today?
Today, I recorded my first commute to my Android powered G1. I wrote an application that allows me to press a button to start/stop recording my commute. It notes the start and end time, plus saves GPS data about waypoints along the way. A future version of that app will upload the data to a web server, but for now it just collects it into a database.
Any other commuters out there be interested in something like this?
P.S. “Long commute” is a very relative term. A half hour might be long for some people, and a 3 hour commute might be long for others.
I used to have a ~1.5 hour commute each way, by train and walking. I now live much closer to work and have to say if anyone can optimize their commute by simply moving, they could be a lot happier. Obviously that isn’t for everyone, but the time and energy people put simply getting to work is amazing.
That said.. this sounds like a pretty sweet app. A bonus feature (maybe part of phase 5) might be to suggest carpool-mates for people who travel the same route.
My commute was 96 miles one way. My company now moved a little bit closer, I now have a 87 mile one way commute. It sucks!! In June 1, 2006 I purchased my car new. Needless to say, it now has 164,000 on the clock. Gotta love my Corolla. Hardwick, MA to Hamden, Ct.
You know this project (and data) sounds like something worth turning graphically – like Edward Tufte.
I look forward to the poster!
Joyce