Application of Convolutional Neuronal Networks in Customer Base Analysis
(A2024-119378)

AUTHORS

Shahrzad Kurbiel, University Duisburg-Essen

ABSTRACT

Customer lifetime value is a useful metric in customer relationship marketing, offering insights into distinct customer segments. However, predicting future purchase behavior poses a significant challenge in calculating customer lifetime value, particularly in noncontractual business relationships. In recent decades, various approaches have emerged to predict future customer activities. In the literature of customer lifetime value, probabilistic models represent the most widely used approach, which describe a purchase process based on predefined assumptions. While these assumptions may reflect the inherent features of a purchase process, their applicability cannot be generalized for all purchase processes. To address this, the current study employs the deep learning approach. At its core, a convolutional neuronal network is designed and evaluated on two real-world data sets. Depending on the data, the evaluation of the model shows a better forecast compared to the benchmark models.