The following project aims to develop a Customer Segmentation for a fictional insurance company and to provide a clear and proven profiling of the 10296 customers in the data set, in order to be delivered to the Marketing Department for further development.
First of all, we started by exploring the data set, getting a better grasp of its behaviour and possible incoherences. Then, it continued by pre-processing the given data: finding and removing the outliers and filling the missing values. After these steps, we looked over all variables and we discussed the goal of our clustering, defining which segmentations would make sense to do.
The first segmentation clusters were achieved by using the combination of the K-means algorithm and the hierarchical clustering. The clusters from the second segmentation were achieved by using the K-Prototypes. At the end, we got the final clusters crossing the results from both segmentations. Finally, we elaborated succinct marketing approaches hinged on our final clusters.