Sunday, July 26, 2009

A quantitative study of two options for the commute to work

Greetings all! While I know that this post will not break the trend of west-coast heavy postings, I felt the information was interesting enough to share.

Introduction: I have two commuting options to work (not counting driving), the Glen Park BART Genenbus, and the Marina/Pac. Heights Genenbus.

Glen Park BART Genenbus Pros:
  • Picks up every 10 minutes
  • Short ride on the Genenbus (relative to Pac. Heights Genenbus)
  • I get to ride the cable car to work (aka, awesome)
Glen Park BART Genenbus Cons:
  • Requires three modes of transportation (cable car, BART, Genenbus), compared to 2 for the Pac. Heights bus
  • Seems to take slightly longer than the Pac. Heights bus based on previous experience

Marina/Pac. Heights Genenbus Pros:
  • The pickup is a short walk/Muni ride from the apartment
  • Previously observed relative consistency in arrival time at work
Marina/Pac. Heights Genenbus Cons:
  • Only picks up every half hour or so
  • Greater distance to travel on the freeway (greater opportunity for being caught in traffic)

I decided to track my commute, to determine which option was most favorable.

Methods and Materials: Time was tracked with minute precision based on AT&T Wireless time. Thirty commutes were monitored on the Glen Park Genenbus route. Said route entailed: wait for the cable car, ride the cable car, walk to/wait for BART, ride BART, walk to/wait for Genenbus, ride Genenbus. Sixteen commutes were monitored on the Marina/Pac. Heights Genenbus route. Said route entailed: walk to the Genenbus stop (OR, wait for Muni, ride Muni to Genenbus stop), wait for Genenbus, ride Genenbus.

Every effort was made to cover a reasonable range of arrival times on the Glen Park BART Genenbus route. All Marina/Pac. Heights Genenbus data is based on the 8:30AM-arrival bus (note: the 8:00AM and 9:00AM-arrival busses are only used on the rarest of occasions).

Results and Discussion: The mean commute time on the Glen Park BART Genenbus was 56min:20sec, with a standard deviation of 6min:20sec. The minimum and maximum commute times were 46min and 1hr:09min, respectively. There was essentially no dependence of the trip length on the departure time (as seen in Figure 1).

Figure 1.

The majority of the variation in total trip time was caused by the wait for the cable car (up to 14min!!!) and the Genenbus ride (data not shown). Both the wait for BART and the BART ride had extremely low variability (as would be expected).

The mean commute time on the Marina/Pac. Heights Genenbus was 53min:41sec, with a standard deviation of 2min:06sec. Total trip times ranged from 50min to 57min. It was found that walking was consistently faster than taking the Muni (mean of 17min versus 20min from apartment departure until boarding Genenbus, respectively).

Conclusions: The ideal arrival time at work would be 8:30AM. Based on that, cumulative probabilities of arrival are shown in Figure 2, for both the Glen Park BART and the Marina/Pac. Heights Genbuses. By even 8:25AM, there is a greater than 90% probability of arrival on either the Marina/Pac. Heights bus or the 7:20AM Glen Park BART bus.

Figure 2.

This data supports the Marina/Pac. Heights Genenbus as the most reliable and time-efficient option for on-time arrival at work. Future work will explore the benefits and costs of driving.

4 comments:

Kristen said...

wow Ariel, there are no words...what can I say except that I am proud to say you were my first homework partner at Cornell (even if you were assigned!)

J. Bro said...
This comment has been removed by the author.
J. Bro said...

Blogger J. Bro said...

Your data for time dependence may correlate to a higher order function. In fact, i believe that type of distribution looks very text book to me. I will look through my Stetinger notes after work to see If I can find it.

Also, what is the time/cost analysis. BART is not free. So, the commute will not only fail to get you to work on time 5% of the time, but it will cost you more, too.

:)

Previous post deleted for spelling error.

Ariel said...

Jason- thank you for the follow up. I would be interested to see what distribution fits the "time dependence."

As to the time/cost analysis, perhaps it would be better named commute time and pro/con analysis. I already know that driving would be the least desirable option, fiscally speaking.