Edinburgh, 28-30 August, 2019

Artificial intelligence in credit processes: CRIF at the “Credit Scoring and Credit Control Conference”

Now in its 16th edition, the international “Credit Scoring and Credit Control Conference” organized by the Credit Research Centre at the University of Edinburgh Business School is Europe’s premier conference for credit scoring and related topics, which every year plays host to over 100 presentations over three days.

At this year’s edition, the CRIF delegation presented two original contributions on the innovative topic of applying artificial intelligence to credit processes.

Enrico Bagli, Data Scientist at CRIF, gave a talk on “Analyzing prepayment and default under changing credit market conditions for SMEs by applying advanced analytics on Credit Bureau data”. CRIF, in collaboration with Banca Montepaschi di Siena and the University of Florence, promoted a study on the development and adoption of a model able to assess the profitability of loans given a certain amortization rate. The profitability of an individual loan or loan portfolio was measured by applying a Random Net Present Value (RNPV) model. Over time, the intensity of these phenomena depends on the varying economic conditions, implemented using the “Markov chain” method.

In the second presentation, “Explanations of Machine Learning Predictions: A Mandatory Step for its Application to Operational Processes”, Alessandro Poluzzi, Project Leader at CRIF, introduced the methodology development plan that the company is implementing for the adoption of new machine learning and artificial intelligence technologies, maintaining requirements related to explainability and impartiality in the context of data-driven decisioning. Giorgio Visani - who is currently studying for a doctorate at the University of Bologna and funded by CRIF - followed, looking in detail at a credit risk application consisting of the use of LIME techniques. These enable a local explanation of the results of a model to be provided, highlighting the quantified benefits and potential focal points.