• Create two separate customer segmentations (using clustering) to split them into 3-5 clusters based on two factors:

  • Demographics
  • Banking Behaviour (savings accounts, debt, credit, transaction numbers etc.)

• Visualize the created clusters using radar charts and compare them agains each other

• Visualize segmentations using scatter plot (PCA since we have so many features (dimensions) and we want to plot our observations on a 2D plane)

• Visualize in 2D how our clusters are evolving in each iteration of KMeans (for at least 20 iterations) (using loops)

• Data Wrangling
• Data Visualization
• Data Preparation and Feature Engineering
• Dimensionality Reduction
• Unsupervised Learning

The project was broken down into two segments since our goal is clustering, based on two major criteria demographics and banking behaviour. Initially we used demogrpahics as our minimal viable product (MVP) and it created the general structure and called this part one. Then used the designed workflow from MVP to design our part two of this project which is Banking Behaviour.