Hello,
We’ve noticed that not all planes appear in the first position data. For example, in KMEM, plane id AAL1027 does not appear in the corresponding first position data but AAL1037 does. In fact, at KMEM, more than half the planes in the train labels don’t appear in first position. This is a bit confusing because we don’t see any reason why it wasn’t recorded. Is this by design?
Hi @surajraj99 ,
Thanks for your question. The first position dataset only contains flights arriving at the airport. (That explains why the data exists for about half of the GUFIs present in other datasets; the other half are departing flights.) This is by design insofar as it is true to how the data is available for a particular airport in real-time. I agree it’s not what one would expect; I’ll add a note to the problem description to make that more clear for others!
I see. Thanks for clarifying. One additional question. For the example of AAL1027, it’s a plane that is departing from KMEM. But for that to happen, it should have arrived at KMEM at some point right? And under the same plane id of AAL1027 (even if the entire GUFI is not the same). So if it (AAL1027) is departing from KMEM (it should have arrived at some point as well, under a different gufi but same plane id), then it should be in first position? Is there something I’m missing?
Thank you!
Ah, so the first part of the GUFI is not a plane ID that follows an aircraft, but more like a flight number that can be (and usually is given what we see in the data) refreshed at the start of each flight. (As far as we know, there is not a way to identify a specific airplane with the data we have for the competition.)
Update: “there is not a way”. (Is it a typo if it completely changes the meaning of the sentence?)
Sorry, just to confirm, are you saying that there IS a way to identify a specific airplane with the data we have for the competition? Other than gufi? Was doing some research and it seemed like the unique identifier for a plane is the registration number or the tail number.
Apologies, that should have said “As far as we know, there is not a way to identify a specific airplane with the data we have for the competition.”