Industries are using Data Science or Machine Learning in a creative way to improve user experience. One of the enterprises that use Machine Learning successfully is Netflix, but, how do they exactly apply Data Science?
Netflix y Data Science: content customization
Have you ever wondered why your Netflix profile recommends a different show whether it’s you watching or your parents? Or how can they profile each and every one of the 120m users subscribed to Netflix, in 190 countries? The answer lies behind the data they store.
“Traditional entertainment companies don’t have the data we have.”Kelly Uphoff, responsible for Netflix’s Marketing Science & Analytics
This data is their gold mine. A mine to which they apply Machine Learning to analyze and extract value from that data with a clear objective: improving the experience of their users.
“We know at which time of the day you look at contents, how much time you looked at them, we know what you saw before, what you saw after, whether you watched it on your computer, smartphone, tablet, or in a screen, in which profile you saw it. We have a lot of information.”Todd Yellin, Vice-President of Product & Innovation
The “Stranger Things” thing
If you have not watched Stranger Things, you must have heard about it at least. It’s a great example of the application of Machine Learning because the way they offer the previewing of their content changes in function of the user that’s watching.
For example, if your data shows that you like a particular actor, let’s say, Dustin, when logging in you’ll see images of that actor. If what you like is children’s adventures, you’ll see the image of the protagonists. And if you have films or horror shows among your preferences, you’ll probably see a much more terrifying image… don’t believe me? Check it yourself!
Improving the service improves the experience
Another way in which Machine Learning is being applied is that it uses its own historic data to improve the streaming quality service. It is constantly analyzing the viewing time, the device it is being watched from, and removing data traffic congestion in certain networks or making a more efficient geographical redistribution of data in their servers.
Their analysts keep searching for new ways to improve their service, based on creativity and trying to find new solutions to new problems that come up after the great transformation that Netflix has started in the audiovisual industry.
👉 Keep reading: Top 15 techniques to develop creativity in tech
How their suggestion system works: an algorithm with 27000 micro-genres
One of the major assets, and perhaps the one that has brought them more value, is their suggestion system. Their algorithm is based on creating very concrete micro-genres for their contents, up to 27000, some of which are “African-American Social Issue Dramas” or “Classic Musicals based on Children’s Books“. This allows them to customize with great precision the degree of suggestion for the films and series they offer every user.
In 2016 they changed the way in which they analyzed data. Before, they segmented by geography, age and gender, and now they do it by catalogue, functioning globally and generating around 2000 communities of preferences and tastes. Or clusters, as they call them in the data world. Depending on what you see, you’ll belong to one community or another in which different contents will be suggested.
“Black Mirror works for cluster 290 and 56, where people also like Lost or Groundhog Day”States Yellin
From customization to content generation
The last step in the use of advanced algorithms in Netflix has been the creation of their own content: “we use the information we have to create shows, but that’s not what’s most important. We are not looking at information and saying ‘we need some drama with four boys and a girl with powers, and a parallel world called the Upside Down and a monster’”, said Todd.
A curious case is the acclaimed and award-winning Roma, produced by Netflix and barely distributed on cinemas.
There is more Machine Learning than Netflix: Applications in the financial sector
Each time there are fewer industries not implementing Machine Learning, Big Data or Artificial Intelligence algorithms in their models, but that’s not the case of the financial sector, where its applications are diverse:
- Storage of great amounts of unstructured data
- Data analysis for decision-making
- Optimization of credits and loans
- Risk analysis
- Client selection for preauthorized credit
- Default detection
- Process automation
- Automated trading (Robo-advisors and Quant-advisors)
One of the last applications implemented by Banco Santander are Chatbots, the assistants based on Artificial Intelligence that are being used in both smartphones, via Apps, as well as in the web’s online manager.
To sum up
Machine Learning, like all new technologies, has a lot of potential yet to discover. Netflix is just one example of an enterprise knowing how to exploit it and basing their success, mostly, on it. In its applications within the banking sector, there are still many challenges, and the success of its application will be decisive for digital transformation, both in the financial sector as well as with client services.