In order to address the next marketing campaign, we need to identify the clients that are more likely to subscribe. Since it's hard to evaluate the clients one by one, we can determine homogeneous groups of clients and identify the most promising groups.
Starting from the past data, we cluster the clients on the basis of their personal details. Then, given a new client, we identify the most similar group and associate the new client to it. We don't have information about the customer behavior of the new clients, so clustering is based on the personal attributes only.
There are different techniques performing clustering and in this section we use a relevant algorithm that is hierarchical clustering. A parameter of hierarchical clustering is linkage, which is the way of computing the distance between two groups. The main options are: