/* ---- Google Analytics Code Below */

Friday, March 11, 2016

Bayesian in R for Estimating Waiting Times

Nice example of using Bayesian estimation in R.  Relatively simple example by James Dreiss. Developing and using Priors.   Recall just this sort of question for a very different problem.   Includes instructive code and visualization.

" .... Say we're interested in estimating the time we'll have to wait for the arrival of a G train, so we start to collect data on the times between trains. We obtain the first and second day's measurements, and it's not what we expect. For both days there's a fairly short time between trains — only 3 and 4 minutes.

Would you expect this to be an accurate representation of the average time between trains? No, probably not. These are obviously just two data points, and a three minute wait time does not exactly measure up to data on the G train's scheduled service from the MTA. We should of course collect more data, but in these early stages when we have so little of it, is it fair to totally discard the legitimate (as in, empirically backed) belief that the average time between G trains is longer than 3 or 4 minutes? .... " 

No comments: