Banks are an integral part of society for obvious reasons and credit score is an
important metric that can determine the trajectory of one’s life. This score is crucial in determining eligibility for
loans, mortgages, and sometimes even employment opportunities.
Our team is at the forefront, utilizing machine learning and advanced data processing
techniques to anticipate credit defaults.
Our methodology includes performing an ablation study to evaluate the effectiveness, as assessed by AUC, of a machine
learning model trained on a dataset with processed features in comparison to two benchmarks: the identical model
architecture trained on unprocessed data and a traditional credit score approach.
This comparative analysis aims to elucidate the impact of data processing techniques on improving the predictive
capabilities of machine learning models in predicting credit defaults. Our objective is to contribute to the
comprehension of how advanced data processing may result in more precise risk assessments within the realm of financial
lending.