American Express payment company have organised a credit default prediction competition to create a better customer experience for cardholders by building a top machine learning solution that challenges the current model in production.
4875 teams from different countries have participated in this contest and suggested their ways to predict credit default. Over the three months participants had access to training, validation, and testing datasets including time-series behavioral data and anonymized customer profile information. Teams сould use any technique to create the most powerful model, from creating features to using the data in a more organic way within a model.
We spoke to Daniil Orel, a 3-year student who studies Computer Science at NU SEDS, to learn more about this contest. “I often participate in hackathons and datathons – hackathons for analysts. About AmEx competition, I have learnt from DSML.kz chat. I saw the task and realised it was close to what I had done before. The task was to use anonymised client factors to predict the default of this client over six months. We took silver and 209th place among all participants. However, this competition gave me a lot of experience in the practical application of ML-algorithms”, shared Danіil about his experience.
We wish our students further success in their endeavours!
