I have a dataset of an insurance company for my data science class project. My ultimate business objective in this project is to sell more policies to existing customers/customer segments.
Firstly, I want to cluster my customers through k-mean model with RFM scores then use apriori algorithm to find association rules among this clusters. Later, I can find which customer/customer segments I can sell more product/s. Yet my teacher want me to test my prediction and said that since the policies are repeated every year, you can not split your data in terms of last 3 months is test data-set and the rest of the 9 months is train data or etc. To sum up, he wants me to test my prediction in more accurate way. How can i test my prediction in this specific case?
My data set includes not much demografics info about customers such as age, income, education, or etc. Then I want to use RFM scores since I know the customers' all purchasing records. Columns include what policy type they purchase, when they purchase, which company they purchase with, the pricing of the purchase, which region they purchase in, telephone numbers, adresses, mail adress etc.
Insurance types are life, car, traffic, residence insurance, fire etc.
