Customer Segmentation of Credit Cards

Project information

In this project, the marketing team of a fictional bank had 6 months of transactional data of customers using credit cards. They wanted to launch a targeted marketing ad campaign tailored to a specific group of customers. In order for the ad campaign to be successful, the customers have to be divided into 4 distinct groups :
Transactors, New Customers, Revolvers, and VIP/Prime

This was a clustering problem, so I used the k-means algorithm. I also used the Elbow method to find the optimal number of clusters. And finally, ran PCA to reduce the 17 dimensions into 2, so the clusters could be eaily visualized in a 2D graph/image.