Data Science in Finance Conference
On 20 April, I participated in a panel discussion on the impact of data science in the financial sector (Data Science in Finance Conference). It is impressive to see what results have been achieved by the banks. Above all, I appreciated the openness and readiness to share knowledge with each other, not only between banks but also between supervisor and banks.
In the morning, Joost van der Burgt, chief innovation officer Toezicht at DNB, elaborated on the use of data science, the role of regulators and self-regulation in the financial industry. The degree of regulation is highly contextual: it depends on the context in which the algorithms are used.
Fairness and discrimination are inversely related to each other
I also followed the presentation by Joris Krijger and Daan Knoope on the topic of Fair Machine Learning. I could summarise as follows: ‘You can’t have your cake and eat it too’. This also goes for machine learning models. Good accuracy can be achieved if the algorithm is able to discriminate between groups. Otherwise, predictions are made without telling us if the prediction is correct or not. At the same time, good discriminatory power will negatively impact fairness. Daan and Joris showed how fair discrimination can be achieved if we are intentional about it: in a justified, legal and moral way.