Predicting credit default

In the financial services industry, one of the major sources of losing out on revenues is the default of certain customers. However, a very small percentage of the total customers default. Hence, this becomes a problem of classification and, more importantly, identifying rare events.

In this case study, we will analyze a dataset that tracks certain key attributes of a customer at a given point in time and tries to predict whether the customer is likely to default.

Let's consider the way in which you might operationalize the predictions from the model we build. Businesses might want to have a special focus on the customers who are more likely to default—potentially giving them alternative payment options or  a way to reduce the credit limit, and so on.