Transcript
Now that we know, from a previous episode, what Big Data is, we will discover today, what you can do with that data. Machine learning is the buzz word here.
Hi Ben, welcome back to our studio.
Hello.
In last episodes we talked about Big Data. Collection of Big Data.
Now we're going to do something with the data. And that's machine learning.
Yes, exactly. So, machine learning is actually the...
Let's call it an art. The art of handling that data and extracting meaning from it.
Everybody knows machine learning nowadays, because it has become more and more popular, the last years. Especially, it's widely used in marketing, for example. In e-commerce, people all know the fact that when we view a product online, that we get to see other products, that people eventually bought or also bought or also liked or also viewed. And, behind the scenes, that's machine learning in action. It's looking at the number of times people have viewed certain items.
And actually, it's just advanced counting. It's seeing: okay, if a person has seen a certain item, in combination with another item and that happens a lot of times, these items, they probably belong together or are alike.
But is it then just buying a machine learning computer, switching it on and you get results like that?
It's relatively easy to do, actually. But it's really hard to master. The starting point is getting to know the different kinds of models. Knowing which model you need to use in which situation.
So, for example, if you want to predict how many people will be coming to your event this year, based on your last events, you typically look at a regression model, which can predict a distinct value or a number.
If you want to categorize people into which types of events they might like, you're looking at classification models, for example.
So, you really need to get to know which type of model you need to use. And once you know that, it's just a matter of calling a library and putting your data in there. And selecting the right model and your output will come out.
Of course, it's harder than that, because you have to do your feature engineering as well. Make sure you get the right type of data in the right form, in there. And that's what makes it really hard, but, as with everything, the first 80% is relatively simple. And the last 20% becomes really, really hard.
You already mentioned predicting how many visitors will come to your next event, based on previous events. Are there other applications for machine learning in the event industry?
You can look at marketing: selecting the right tickets for the right people. But also, the right marketing channel for the right people.
What do you mean with that?
You can look at how many people respond to a certain type of marketing channel, in some way. Are people more responsive to e-mail marketing or to physical mail marketing? Or even telephone calls?
Or WhatsApp, these days.
Yes, exactly. And you can look at that, based on your previous data and make sure that you get the highest response possible. Based on that.
Yes, without spamming them.
Yes, exactly. Privacy is a big concern, nowadays. In marketing. For the good.
But that's also something you can use machine learning for. Knowing when not to spam a customer anymore, or when he won't respond anymore.
So, it doesn't always have to be a bad thing.
Are there also applications possible during an event, for example?
Yes, you can track where people are going and steer that. And make prediction models on the busiest spots in an exhibition, for example. And steer people into quieter spots and make an optimal route for them within your exhibition, for example.
An intelligent crowd management platform.
Yes, exactly.
You can start monitoring what people are looking at. You can use that, from your previous events, to lay your hot-spots further apart. So, you have a less crowd intense area.
So, there are lots of possibilities, actually. And usually, it comes to the fact that you get to know a customer more. And that you personalize things more, for them.
If I hear you talking about machine learning, it sounds to me that it are really big projects, where you need a lot of budget to implement. So, it might only be for the big festivals, like a Tomorrowland or something like that.
But are there also possibilities for small events, small companies, to start working with Big Data?
With machine learning.
With machine learning, indeed.
Yes, but it's funny that you mention Big Data, because often people confuse the two. Or think that they go hand in hand. But you don't need a bunch of data, to get started with machine learning. It's already possible on a relatively small set of data.
We've done machine learning models based on not more than a hundred events. For our training data. So, you don't necessarily need big budgets.
Also, the software we use, is usually from academical sources. And is open source and free to use.
So, free to use.
Of course, the people that are working with it, you have to pay.
But it can start relatively small. So, it's not really that hard to do. It's just a matter of selecting the right models and implementing them.
I won't say it's something every developer can do, but every developer can easily get started with it. And, of course, if you want to elaborate further, and make your models better and better, you have to hire some expertise. But, just getting started, it's, even for a
smaller company, it's really useful.
Okay, I see a lot of opportunities for the event industry, so I think we should embrace these technologies, as an industry.
Ben, thank you very much for coming over and explaining to us.
No problem.
And you at home, thank you for watching our show. I hope to see you next time.
