How Late Will You Really Be?

For fans of FareCast, there is a cool new site called DelayCast that’s just gone into beta.

Type in the airport codes for your departure and arrival cities and the date and site come back with predictions about the probability of cancellation and delay for different airlines serving the route. For example, here are the results for a trip from Chicago (ord) to Hartford (bdl):

And in true Super Crunching fashion, DelayCast not only predicts but gives the 90 percent confidence interval for the prediction. The site’s F.A.Q. explains why their predictions are better than the government’s on-time statistics.

Now what I’d like to see is a DelayCast for other parts of my life.

In particular, I’d like to have a statistical algorithm for accurately “translating” the delay predictions of pilots. My strong sense is that when a pilot comes on the intercom and tells you that the plane will be taking off in 10 minutes, the passengers should expect the plane to leave in 15 or 20 minutes. (The true distribution is skewed right and a good Bayesian would also increase the probability that the flight would be canceled).

Instead of trying to get pilots to speak more honestly about delays, I think we’re more likely to be able to produce a translation program:

When pilot says X, passengers should hear Y.

You can help create such a program. For the next week if you’re traveling by air, please pay attention to all the predicted delays and compare them to the actual delays — and post your numbers as a comment to this post. (I’m a bit scared that people will disproportionately report when they are frustrated by an unexpectedly long delay — so the data is likely to be unrepresentative.)

But if we get enough data, I’ll crunch some numbers. The exercise underscores the needless (or maybe I should say self-interested) bias in many human delay predictions.

When a home contractor says your kitchen renovation will be done in two weeks, you’d be foolish to start sending dinner invitations. But it would be nice to know whether two weeks means three weeks or two months.


My experience with air travel has been that there is so much slop in the published timetables that a flight can be significantly late leaving and still arrive on time.

The most spectacular example of this was a cross-country flight that left Newark nearly an hour late and still arrived in San Francisco 15 minutes early. How about that?!?

I've also found that the larger the airport is, the wider the negative difference between the announced and actual delay. If I'm in my local airport (all of two runways and 30 gates), 10 minutes is 10 minutes. In New York, Chicago and Atlanta, 10 minutes is an hour.

Still, it's better than taking off on time and circling the destination airport...

And even better, the job I moved to has kept my flying to a minimum. 2005 was the first year in nearly ten that I drove more miles than I flew.


Amazon (where I always opt for free shipping for books) seems always to give me a package arrival prediction date that is much later than the actual arrival date. Is this (a) because they want me to remember them as faster than they are, (b) because they want me to opt for paid shipping instead of free, (c) because of a sloppy computer program, or (d) something else?


I work in Statistical Analysis (weird that i read this blog, right?) and i know i'll tell people something is going to take me 2 days or an hour or whatever knowing that it will take me longer than i think it will, and i'll even usually go over what i told the investigators in the first place

Just last week, Midwest airlines said 2 hrs and 30 minute delay and it was closer to 3 hr, i'd say 2hrs 50 minutes. Pretty good, actually.


One thing I've noticed that is specific to Southwest Airlines (I fly many airlines) is that if you're flight is delayed, and they say the new departure time is 7:15, you will begin boarding at 7:15 and will actually depart 20 minutes later - like clockwork!


We were told we would be delayed 10 minutes waiting for some additional passengers, and we would still arrive on time. After 10 minutes, we were told it would be a few more minutes until they showed up, and we'd be close to an on-time arrival. When the passengers finally arrived, we sat at the gate for another 15 minutes without explanation before pulling away. We were about 30 minutes late arriving.


I'll start. Just this morning I was told that we'd be delayed 15 minutes taking off. Surprisingly, we actually only waited 15 minutes, and we arrived on time for our destination.


Have we not yet reached the saturation point for obsessing about time and timeliness? Imagine what life was like before clocks and timepieces had been invented. Back when "on time" meant the right day, or even as far as "in the morning" as opposed to "in the afternoon." When things got done because of how long they took to do, not because somebody computed, then time-tabled, then somebody else crunched the numbers and translated.

Here are my numbers: The other day I took a flight from Kansas City, MO to Portland, OR. The pilot said we were going to take off, and we did. In fact, we got there the same day we left. Not only that, but every member of my party was alive and well, including the oxen.


I think people in general are horrible at estimating time. My rule of thumb is however much time I expect it will take me to do something I add about 10-30 minutes depending on the task or errand. When I am forced to be on the waiting end (i.e. girlfriend getting ready and says it'll take her 20 minutes) I just double it. When dealing with a commercial business or service all bets are off. The time you are given could be off by 10 minutes to 10 weeks.


Are you implying that pilots give inaccurate predictions about expected take-off and landing times? That's unheard of!


I flew today on JetBlue from Boston to DC. We were on time, and in fact 20 minutes early!

Alex F

I'm not sure how useful confidence intervals are on probabilities. I'm pretty sure expected utility theory tells me I only need to look at the mean...


Today on a flight from Chicago to JFK on Jet Blue we were informed that we would have a 5 to 15 minute delay. Well, after 5 minutes, it was announced that we would be delayed 3 hours. We ended up leaving and arriving 2 hours late.

And the tvs didn't work.


Flying from Phoenix to Chicago today on AA (one of the few flights that got out on a MD-80) we had to circle due to air traffic conditions at ORD. When we were finally given the all-clear for final approach, the captain said we would be on the ground in 20 minutes. I rolled my eyes and noted the time. 30 minutes later we touched down. I wish I could get a frequent flyer credit every time the pilots were off on their predictions. I'd never pay for a ticket again!

By the way - the delaycast web site seems very cool. Thanks for the tip!


Today on Southwest from San Diego to Austin.

Original Departure scheduled time: 3:30pm
Delay announced ~ noon: 4:10pm
Further delay announced ~ 3:00 : 4:30pm
Further delay announced ~ 3:30: 4:35pm
Actually pulled out from the gate at 4:45pm.


Departure delay isn't all that interesting. What I want to know is if we're going to arrive on time. I believe that they have some room for departure delay, when they first tell you they arrival time. Most of my flights that depart 15 minutes late, still arrive on time.

Jon S

My 7:05p flight from Boston to Chicago last Sunday was reported to be delayed until 9:45p. It was then announced 10:45. Then 11:30. The plane took off at midnight.


there's also a law of bus delays- when a bus is late, and there are too many people to board, the driver will always say "there's another one coming right behind me"- I used to think this was evil, but realized it's a survival technique- not wise to not indulge an angry mob


If a bus is delayed, and the next bus is not, it follows that the interval between this bus and the next bus will be smaller than normal. In other words, the very fact that the bus is delayed automatically implies the next one is coming up right behind it.

Also, once a bus is delayed, it's going to have to pick up more passengers, delaying it even further (and reducing the interval to the next bus).


Re: Mike's Comments-

While I agree that Americans are typically obsessed with timeliness, I don't think the problem is new by any means. And if you're looking to go back before timepieces were invented, then good luck. According to Wikipedia (caveat: yeah, it's Wikipedia), the earliest sun dials date back to the Egyptians. My other question is this: are you a banker, carpenter, or farmer?


Following up on #16 & 17: If bus 1 is late, it is, indeed, sometimes the case that bus 2 will NOT be, and thus will be closer behind it; however, sometimes the same thing which caused bus 1 to be late (such as snarled traffic), will ALSO make bus 2 late (either just as late, or even moreso).

So the frequent bus-driver's promise, "There's another bus right behind me," will not always be true. In at least some of these cases, said driver will be aware of this, but says it anyway. I agree with Frankenduf that this is a survival technique; but ultimately it does not help the hapless rider who can't get onto a "sardined" bus, because s/he cannot trust that the driver's promise is true.

What WOULD help riders, and give them a promise they CAN trust, is some sort of status display at bus-stops. With modern technology and GPS tracking, this should not be difficult. It would be an expense, yes, but perhaps one that riders could pitch in for voluntarily. The most problematic part of it would be the chance of the displays being vandalized, and this may well be a hurdle that's difficult to overcome.