A topic designed by Warwick Business School requires students to enhance loan portfolio management by utilising cluster analysis to group borrowers with similar characteristics, enabling personalised loan products, targeted marketing strategies, and a better customer support process to serve the unique needs of each segment through cluster analysis based on the provided data
I approach this problem by eliminating the irrelevant data based on my domain knowledge about banking industry. Then I remove the outliers, which are extremely sensitive to the models and randomly sampling the dataset. Also, there are some multicollinearity among the variables in the dataset, which might affect the final result, therefore, I conduct principal component analysis and factor analysis to transform the original variables into the new ones. Finally, I will do cluster analysis on the transformed variables to generate the result.
The coding part has been provided in Code.Rmd and detailed analysis has been provided in Cluster Analysis.pdf file.