DAIM: Inside The Algorithm | Network Forecasting & Data Science Leadership with Dr Judit Guimera Busquets

A very warm welcome, Judit, to the podcast.

Thank you ever so much, uh, for joining us.

Thanks for having me, um, as well, Jeremy.

Um, pleasure to be here.

[upbeat music] I want to crack on with our sort of first, um, question in, on your, uh, research, which was around, around air traffic, uh, forecasting in, obviously, in aviation.

So before we get in the, into the detail technically, I'd like, you know, for listeners who maybe haven't, uh, thought about this problem before and why it's a problem, can you just set the scene for us? Can you, can you sort of explain what, what makes forecasting, um, demand forecasting I suppose we're talking about here, uh, in a network, you know, different and quite tricky from other problems?

I think when people think about forecasting, they think about, like, predicting kind of like, you know, sales for a single product or estimating passengers', um, demand for kind of like a isolated standard, standalone route.

Um, but in air transport, in an air transport network, the system is a complex and, and living system where kind of like everything is connected to everything else.

Um, so I think what, what makes the, the forecasting within a network kind of like framework or like setting, um, different, it comes down to, to, for example, like the domino effect or kind of like the cascade effect.

So, like, traditional forecasting, um, of a product, you know, where you have, like, a change of one of the items rarely will break the rest of your kind of like catalog, right? But in air traffic forecasting, a single change of, um, a, a single change of, like, the network, so, like, maybe an airport pair, um, connecting where before there wasn't a route there, um, or, or, you know, like, uh, two airports that are connected now, you know, get removed that flight that was available, kind of like cascades through the entire system.

Right.

Um, so that kind of like effect, knock-on effect on everything else is what, um, is one of the challenges, um, and one of the differences, um, between kind of like, yeah, um, between, like, a single product, like, forecasting or, or like, you know, a single route forecasting than the network itself.

So this isn't then like a, a, a sort of standard forecasting problem at all in this respect.

If I was, if I was trying to forecast how many, um, cars of particular model or, uh, manufacturer I was gonna sell in the next, um, six months, in the next year, and I'd been selling that car for absolutely, you know, for many years-

Yeah

... um, then, then I, I would just be using, you know, previous sales data-

Right

... maybe something cyclical, maybe something periodic based on the, um, the time of year, the time of the month maybe, and then some, because it's quite an expensive purchase, maybe some economic sort of indicators would all sort of go into the mix and go into, into that, that forecast approach, and I'd end up with something that was, you know, probably reasonably likely to, to, to, to, to give me a, a, a decent outcome.

But what you're saying, and I think this is really fascinating, in, in, in the world of aviation, is that when you go back in time, you look at the sort of previous data, you're not actually looking at the same world as the one you're about, about to enter.

You're looking at a different world because it's, it's differently connected.

Is that, is that, is that, is that the, um, is that the challenge?

Yeah.

Your, your kind of like map of flight or kind of like your schedule, um, is completely different.

It kind of like evolves and changes over time.

Um, so, so yeah.