"You shall know your customers by their transactions – the effectiveness of natural language processing inspired representation learning for customer segmentation"
29 March 2021, 12:30-13:30 CET
Financial institutions have long had detailed information about their customers' transactions. Since 2018, all banks in the European Union have been required by a European directive to share this data with third parties upon the request of the customer. The structure of bank transactions is similar to language in that both are constituted by a categorical sequence of words. Inspired by Natural Language Processing, we develop a model that learns a general profile of each customer from such data. This general and stable representation of a customer can be used for transfer learning. It improves serving customers with relevant information across a variety of domains such as insurance recommendations and tips for increasing savings. The uniqueness of our approach lies in developing an effective customer representation learning method that can be used in downstream transfer learning tasks. It significantly outperforms traditional approaches such as sociodemographic based customer profiling.
Florian is Assistant Professor at the Frankfurt School of Finance & Management. He has completed his undergraduate degree in Economics and Philosophy in the UK. His PhD at Cambridge University was a comparison of three approaches to explanation in management research, focusing on causal inference and the construction of explanatory frameworks. Florian's main research interest lies in decision making under uncertainty and the application of machine learning to managerial and organizational problems. He enjoys combining machine learning with experimental research.
The talks will be held online. If you would like to participate, please fill this form.
For more information, write to firstname.lastname@example.org.