We talk to Stuart Barker, Lead Data Scientist, about data science at Hargreaves Lansdown, who are a 2020 sponsor. Data science is about building intelligence and Hargreaves Lansdown uses this to gain new insights every second. The data science team started small, one person in fact, and are now an established team of five that tackle a wide range of business problems. 

Tell us about yourself

My career path in Data Science isn’t very conventional. I left university to work as a Marine Biologist in places like Bermuda and the Philippines. This job sparked my interest in data and I moved to Thames Water where we set up a data science capability to help predict leakage and deploy a drone team to help maintain assets. I then moved to John Lewis and Waitrose where I focussed on forecasting sales before starting at Hargreaves Lansdown two years ago.

I have worked on a wide variety of problems in my various roles, requiring all sorts of different types of data. But one thing remains relatively static – the solving. From predicting leaks, to predicting sales of bread, the framework stays the same. I have learnt many things from these varied industries but one aspect always sticks out: keep things simple.

Always start with the most simple and explainable solutions and iterate from there.

Keep Things Simple

The best answer is the easiest and most explainable one. As an example, take the following problem: When a supermarket sees an increase in footfall, cashiers are often caught out and not all the tills are open resulting in a long queues for the customers. There are two possible ways to solve this:

  1. Build a capability that analyses billions of rows of data and predicts every 5minutes how many tills should be operating.
  2. Ask the floor staff to keep an eye on footfall and empower them to make the decision there and then to open a till.

The first solution will require a potentially complicated algorithm and the second requires human intervention. For me, the second solution wins every time because it is simple, explainable, and quick. Of course, there are many reasons why building a predictive capability is key (and I will touch on these next) but when picking up these problems, always start with the most simple and explainable solutions and iterate from there.

Data science is focused around making intelligent decisions or gaining knowledge when everything is noisy and uncertain.

What we do at Hargreaves Lansdown

The title of “data scientist” is a relatively recent invention although the problems data scientists have been tasked with solving have always been around. As we see it there are two fundamental problems that data science attempts to solve:

  • How can we make intelligent decisions for the future in an uncertain world?
  • Data is noisy and chaotic. How do we get any knowledge from it?

These are expressed in the three (or however many there are now) Vs of Big Data (volume, variety and velocity) and are addressed by all of the buzzwords we know and love: AI, machine learning, predictive analytics, and so on. However, beneath all of this jargon data science is focused around making intelligent decisions or gaining knowledge when everything is noisy and uncertain.

Hargreaves Lansdown’s Feature Library

To keep things simple we see data science at HL as the creation and maintenance of a library of features. Features provide us with valuable information about our clients and the business. Similar to an actual library we group our features into the following genres:

  • Intelligent features that predict a specific outcome (e.g. how many calls are we going to receive tomorrow at 9am).
  • Knowledgeable features that have been derived from the data (e.g. when do we receive the most inbound phone calls).

These features enable us to answer actual business problems such as:

  • How many new colleagues do we need to recruit for our helpdesk to cope with the growth of the business?
  • How do we make sure our investment in marketing is used most effectively?
  • Are there patterns in how our roughly 1 million clients behave that could help us make things easier or provide a more personal experience?

These problems, big and small, can be addressed by expanding and maintaining this feature library. This combined with a evolving technology stack and an increased client focus on data and ethics enables us to do the right thing for our clients.

Thanks Stuart, it was great to e-meet you and hear about what you do at Hargreaves Lansdown.

Buy Your Ticket