In 2011, Mr Mike Anderson set up Nuroko, a company which makes use of advanced machine learning techniques to unlock value from data. The McKinsey alumni, who holds a Double First in Mathematics and Economics from Cambridge University and once represented the United Kingdom at the International Olympiad in Informatics, talks about the convergence of math, technical skills and business acumen in data science and the opportunities for infocomm professionals in this space.

You have been described as a leading data scientist in the community. What is data science?
I would say it is about using scientific methods to create value with data. You make observations about what is happening in the world, form hypotheses, maybe do some experiments, carry out analysis to understand what the data is telling you, and then come to conclusions. If you are doing a good job as a data scientist, the conclusion should help you make an important positive difference to the business.

What makes data science an interesting area for infocomm professionals to be in?
As a data scientist, you are closer to business impact and you have a great opportunity to solve lots of different problems. You can use your technical skills in a way that creates real value for the business. For example, if you can improve inventory management, it could be worth millions of dollars to the company.
Businesses are waking up to the extent that data is valuable to them, and are willing to pay for skillsets to help them analyse this data. There is a lot of demand for data science expertise right now.

How did you start out in IT and what led you to data science?
I started programming at the age of eight when my dad bought me a computer and when I was 16, I took on part-time work during the summer break assembling PCs and setting up computer networks for people. But my first “real” IT job was when I did an internship with Accenture in 1998. It turned out to be a very good experience, and gave me a lot of confidence in what I was doing.

At one point, I was working at the London Stock Exchange and there was a system outage. We had 45 minutes to fix it and avert a crisis, and I got to see how the situation was managed. I went through the database – the log history of what the system was doing before it went down – and I spotted what had gone wrong. This gave me an idea about how we could fix the problem and talked to the team about it. They said, “Yeah, that might work.” And it did.
Going through an experience like that made me realise that the things I learnt about computers and databases were actually very relevant in a real-world crisis situation.  When you spend a lot of time studying in academia, it is sometimes easy to forget how abstract technical concepts translate into the real world. That is why it is always good career advice for students to seek internships so they can learn from real work experience.

In every single business experience that I have had throughout my career, from the early days and the dotcom period up till now, I realise that nearly every business has a data element to it. Information is like the lifeblood of a business. If you understand what the data is telling you, you will be able to make better business decisions. Of course other things must happen as well. The data is only useful if you can turn it into an action and end up doing something different as a result.

What are the important skillsets required for data scientist?
Data science is a complex area that combines several disciplines. You need to have a good understanding of mathematics because ultimately mathematics is the language of data. You also need to know some computer programming because it is often necessary to write customised algorithms to analyse the data, especially when you are looking at very big data sets.

But it is not just about the data or the programming – it is equally important to have a good understanding of the business domain you are working in. You have to clearly identify the business problem that you are trying to solve. Only then can you ask the right questions. This is frequently overlooked by people working in IT and they may waste time on technical things that are not important, so business understanding is really critical.
Data science is a great career move for people who have the mathematical and technical skills, but who are also interested in the business aspects. You have to be curious about business and the data to be a good data scientist. It is great for people with “polymath” tendencies – people who like to mix arts, sciences, economics, languages etc. and are generally curious about many different fields. Everything is connected, and you get a lot of insight from taking a broader perspective.

How can an infocomm professional gain a foothold in data science?
I think it is valuable to be proactive – you can get recognised as a data scientist by practising it in your own company.  Infocomm professionals are already working closely with data in their organisations. It is a great opportunity for them to explore the data and discover some valuable insights. You might find, for example, that the business has too many people working at the wrong time of day to meet customer demand. And if they find something interesting, they can discuss with their business colleagues to see if something can be done to improve the business based on the data. That’s data science. Just do it.

In 2011, you set up Nuroko, which specialises in machine learning software for real-time pattern recognition and predictive analytics. What led you to do this, and can you tell us more about what Nuroko does?
I spent a long time as a strategy consultant with McKinsey – I worked there for around 10 years, with a couple of breaks to complete an MBA and work in software start-ups.

I started Nuroko for a combination of reasons. I wanted to build a tangible product that would make a real difference to the world. I saw the ever-growing opportunities in data science. And I saw a specific opportunity around improved machine learning using new algorithms.

Nuroko creates artificial neural network software, which is a mathematical model of relations between different variables. There are millions of nodes in this model, connected in different ways to try to understand the relationships, similar to how the human brain processes data. We use this to solve complex problems where the data may be quite difficult to understand otherwise. In particular, Nuroko neural networks can detect patterns in data that traditional data analysis methods can’t handle.

Take the example of healthcare, which can have very complex data. The data is uncertain (medical diagnoses are often incorrect), a lot of information can be unreliable (people make mistakes inputting data), the structure of the data is very complex (many different types of data for every patient across many years) and data is often missing, so we need to design methods that can deal with these problems.

Where do you see all these developments with data leading us?
I think in five to ten years’ time, people will be using data pretty much ubiquitously. We will see a convergence of different technologies – machine-to-machine communications, sensor networks, data science, visualisation technologies (which is important if we want humans to understand data), machine learning (which is what I am doing with Nuroko) – all of these things are going to come together to make data available and accessible. We will have a data revolution where people live symbiotically with all this data, and they are going to use it in all kinds of creative, imaginative ways. It’s going to be a very exciting time with a lot of change – on a similar scale to what happened with the Internet.