3

How many, on average, out of 470 people would miss a connecting flight via Dubai?

Calchas
  • 37,112
  • 4
  • 92
  • 148
Jade
  • 41
  • 1
  • 1
    Could you elaborate what exactly you are asking, it is pretty much unclear to many of us. Also provide some background as to why you are asking this question please. – mts Sep 16 '16 at 12:35
  • maybe 10 or 20. Note however that of those 470, not all of them are just connecting in Dubai (ie, you are asking about people flying from X to Sydney, via Dubai); many of them would simply be flying from Dubai (to Sydney). Why do you want to know this? BTW there's an absolutely fabulous hotel (rather hidden) actually inside the airside of dxb, it's awesome (assuming your company or someone else can pay!) if you miss a connection. – Fattie Sep 16 '16 at 20:36
  • 1
    What - why was this put on hold? This is becoming ridiculous. If you can read English, you can understand this question. – Fattie Sep 16 '16 at 20:37
  • @JoeBlow Presumably because OP at one stage says connections on a flight FROM Dubai to Sydney, then later says VIA Dubai, so it's worth clarifying. You can connect from Dubai via Bangkok to Sydney, for example. – Mark Mayo Sep 21 '16 at 14:07

1 Answers1

7

This is a complicated question and predicting missed connection rates has been a topic of intense research in yield management since Beckman's famous paper on no-shows in 1958. At the time, the mean no-show rate was approximately 10% in the United States, but carriers did not distinguish between no shows and missed connections. In 1985 American Airlines estimated a spoilage rate of 15% on sold out flights across its network, but improvements in yield management and the adoption of overbooking had mitigated the economic consequences. (Smith et al., 1992.)

The literature typically elides the distinction between a true no show and a missed connection, at least until the 2000s. A classic introduction to the problem is given in Rothstein, 1985, who also discusses some of the relatively primitive models used at the time to predict no shows, including arising through missed connections, but more recent texts are certainly worth reading if you have the mathematical chops for it; for a flavour, see Subramanian et al., 1999, Lawrence et al., 2003, Garrow & Koppelman 2004, and Neuling et al., 2004. (Free versions of these papers can be found online if your library does not subscribe to these journals.)

In terms of historical averages, the specific answer for a specific flight at Dubai will be proprietary information for the carrier there. But even so, it will vary everyday because a passenger arriving originating in London and flying directly to Dubai may be less likely to be delayed than a passenger routeing himself via several points before he gets to his onward flight in Dubai. Thus the historical profile is not necessarily useful in predicting future behaviour, particularly when optimizing close to the margin of noise.

Without access to those internal data collected by the airline, or knowledge of the yield management policies applied by the airline in question, it is hard to speculate. At a typical hub, however, the missed connections rate is often about ~5%, suggesting that on the order of twenty people would be expected to miss their connection on a sold out 470-seater plane. The airline would be obliged to rebook them.

Calchas
  • 37,112
  • 4
  • 92
  • 148
  • 2
    +1 Epic answer -- and a surprisingly high rate of missed connections! – lambshaanxy Sep 16 '16 at 12:55
  • The only good answer I've ever seen on any SO site. What a shame SO now owns the rights to it. – Fattie Sep 16 '16 at 20:31
  • Say (Calchas) ... I wonder if, on very long haul legs (Dubai-Sydney a great example), there are less misses due to a missed connection: I'm thinking, extremely long combinations are a bit more organized: one might take a risk on a short connection on two short flights within the US say, but you probably wouldn't when flying globally. Perhaps. – Fattie Sep 16 '16 at 20:44
  • @JoeBlow Hard to say, the longer legs can certainly be fed by shorter feeders, perhaps indirectly, and some of those might be infrequent. If you have anyone doing, e.g., SYY-ABZ-LHR-DXB-SYD, it will have a disproportionately high chance of going wrong. – Calchas Sep 16 '16 at 22:13
  • 1
    @JoeBlow Kind words, although they bring your opinion of my other answers into doubt. ;) SO own a perpetual licence to use and publish my answer, they do not own all the rights to it. See http://stackexchange.com/legal article 3. – Calchas Sep 16 '16 at 22:15
  • "... can certainly be fed by shorter feeders" Ah good point, that's true, i forgot about that. Article 3....... (A) "You grant Stack Exchange the perpetual and irrevocable right and license to use..." absolutely any way they want and (B) you can Not have it removed from SO if you happen to want to and (C) you can Not use it as you wish elsewhere (publish a book of your own posts, say) ("Why I write such excellent Answers", perhaps? :) ) if they can impute it was not 100% yours. Tough clause! – Fattie Sep 17 '16 at 04:19