Short description:

In order to be able to offer existing customers further insurance products, it is essential to know for which communication channel (e-mail, letter, portal, etc.) the customers are affine. Before using suitable AI methodologies, a thorough analysis of the relevant systems and the available data is necessary.


As part of a preliminary analysis, all relevant data sources related to channel affinity are considered in detail. This includes, in particular, the insurance company's inventory management system and a data lake in which additional information about existing customers is stored. One focus of the analyses is the contact history, in which all communications between the customer and the insurance company are stored.

Technical description:

From the various data sources, the available features were looked at in detail. Using Jupyter Notebook in conjunction with Python, the features including (if possible) their distribution were visualized and documented and described accordingly. In addition, the features were checked for plausibility.