Predicting customer churn in the fast-Moving consumer goods segment of the retail industry using deep learning

The non-contractual environment, many brands, and substitute products make customer retention relatively tricky in the fast-moving consumer goods market. In addition, there is no such thing as a completely loyal customer, as most buyers purchase from several almost identical brands. If the customer leaves the transaction without notice, the company may need help responding and compensating. Companies should proactively identify potential customers before they leave the deal. Transactional data, readily available in point of sale (POS) systems, provides a wealth of information that can be harnessed to extract customer transactions and analyze their purchase patterns. This offers a robust foundation for predicting and preventing customer churn. This research shows how transactional data features are generated and are essential for predicting customer churn in the fast-moving consumer goods sector of the retail industry. This research presents data concerning the customers of a capillary sales and distribution company in the food industry. We have implemented standard machine learning methods with the available data in this research. However, we have also employed advanced deep-learning techniques to enhance our predictive capabilities. The results and accuracy of these methods, including Convolutional Neural Network (CNN) and Restricted Boltzmann Machine (RBM), have been thoroughly compared, providing a solid basis for our findings.