What’s the most interesting data project you have been involved in recently?
Tough choice I must say, but if I would have to pick one, I would probably say social network analytics. We have been studying this in both churn prediction and fraud detection and found customer behavior to be very social and thus connected. A key challenge however is the design of the network. More specifically, the definition of the nodes, the links, and if needed the weights. E.g., in an insurance fraud detection context the network nodes can be claims, claimant, insured, car, car repair shop, mobile phone, etc. The links can be weighted based upon interaction intensity and time. Building analytical models for these multi-partite networks is a real challenge, but at the same time very exciting. Obviously, it requires a very careful and close collaboration between the data scientist and business user where both can learn a lot from each other!
What was your most important learning during this project?
That the preparation of data is crucial. We all know the GIGO-principle (Garbage In, Garbage Out), and in social network analysis, it’s even more important. You have to closely check all available data; transactional, historical, structured, unstructured, etc. Then look for all deviations and think of how to summarize that in a social network graph. You also have to take into account the operational efficiency of the analytical models you develop. In a setting for the detection of credit card fraud, you have to be able to take a decision in less than 5 seconds. When creating a graph in that case, it makes no sense to take a long period of time as analyzing that period would take at least as long.
Do you see a lot of organizations that do not fully use their customer data and if so, what stops them from doing so?
I do see that a lot indeed and I think that, as always, data is the reason. Despite technical evolutions, we still carry the problems from the past with us with data stored in different places, in several business units, on different platforms and in different formats. In case you’d wait till you have everything sorted, combined and poured in the same format, you would never be able to act quickly. That’s why organizations should start with small analytics projects, using a strictly delimited set of data, and enriching the used models as soon as the data architecture and management mature.
What’s the most important development in customer intelligence in the last 3 years?
That’s the evolution of analytics from a strictly technical and mathematical science into a technology relevant for the business that delivers added value and offers new opportunities.
Why should a (future) data scientist take the masterclass ‘Advanced Analytics in a Big Data World’?
Participants get insights in how to build analytical models using, predictive, descriptive and social network analytics. They will learn how these analytical techniques can be used in different customer intelligence applications (risk management, manufacturing, telco, retail, advertising etc.) and how to ensure the practical application of these techniques to optimize strategic business processes and decision making.
For more information about the masterclass 'Advanced Analytics in a Big Data World' and to register for the course, click here.