Imperial College London Centenary
About Imperial
About ImperialContacts/getting hereAlumniResearchCoursesAbout this site
Select your text size  for this site here: Small Text Normal Text Large Text Extra Large Text

Note: Some of the graphical elements of this site are only visible to browsers that support accepted web standards. The content of this site is, however, accessible to any browser or Internet device.


Next customer please! - the maths of retail banking

Embargoed for time of delivery
09:30 am Friday 7 September 2001

Have you ever been refused a mortgage, a loan or a credit card? Is your credit rating lower than you would like?

If so you might be interested to hear from the mathematicians who provide banks with the tools to rate your financial standing.

At the British Association Festival of Science at Glasgow on Friday 7 September, Professor David Hand of Imperial College will reveal a few of the tools that mathematicians have developed to help banks choose and monitor their customers behaviour.

He uses statistical and machine learning methods to analyse the huge streams of data and to build predictive models of customer behaviour for retail banks.

"Some banks handle a billion transactions every year. You cant analyse data sets of that size by eye - you need advanced statistical tools," Professor Hand said.

"In credit scoring work, we look at data from previous customers, and base our models on how good they turned out to be. Then we can apply these models to the information a new applicant gives on their application form to decide whether they are a good or bad risk.

"Moreover, such analysis can reveal interesting patterns of behaviour. For instance, you might notice that people behave in a certain way before they fall into arrears with their repayments. Or you might spot patterns which have been associated with fraud or stolen cards in the past.

"Well known examples include sudden purchases of many small electrical goods or lots of jewellery - items which are easy to sell on. If a customer hasnt done this before, or suddenly starts this behaviour, it should arouse suspicion."

Computers are vital for retail banking statistical analysis. Professor Hand will discuss how they are revolutionising this work and explain what the computers can do and what this means to customers.

"We use the computer as the key tool. You cant do modern statistics without one. The computer probes data looking for interesting patterns - you tell it what to do and it does it."

Professor Hand will discuss the main classes of methods, including risk assessment, neural networks, and data mining, as well as the development of methods specifically aimed at retail banking. (See notes to editors).

"We are experts on finding patterns in data. Using statistical methods we model customer behaviour. We have developed models which are better and more predictive than the ones currently in use - and a tiny improvement in predictive power can have a big effect on a banks profit," said Professor Hand.

For further information please contact:

Professor David J. Hand
Department of Mathematics, Imperial College
Tel: +44 (020) 7594 8521

Tom Miller
Imperial College Press Office
Tel: +44 (0)20 7594 6704
Mobile: +44 (0)7803 886248

Notes to editors:

  • Risk Assessment is used to see if a customer is a good or bad risk. In credit scoring work, the bank will look at the information on an application form for a potential customer, perhaps combine this with information from a credit bureau, and then decide if they are a good or bad risk.
  • Neural networks transform/combine data in a complicated way - so you can get a much more flexible, sophisticated and accurate model. These and other models are used to predict things such as default risk, probability of switching to another mortgage supplier, probability of taking up an offer, and so on.
  • Data mining is the analysis of very large data sets. very large data sets. Such sets are common in retail banking. The name data mining derives from the analogy of searching for exciting nuggets of information - like gold or diamond mining.

Professor Hands research group develops statistical and machine learning tools in areas such as classification and data mining, and has a special interest in the application of these tools in retail banking, credit scoring, and customer behaviour modelling.

For further information about the group, please look at their website at:

Imperial College of Science, Technology and Medicine is the largest applied science, technology and medicine university institution in the UK. It is consistently rated in the top three UK university institutions for research quality, with one of the largest annual turnovers (UKP339 million in 1999-2000) and research incomes (UKP176 million in 1999-2000). Website: