Published Paper Details:

Abstract: Customer segmentation is an important method both in customer relationship management literature and software since it directly relates to customer satisfaction of the companies. The most common way to separate customers into two distinct groups is to tag a group of customers with a special label. In this paper, company's likely segmented customer data and related statistical data are used to train and test a neural network-based machine learning model, namely Multi-layer Perceptron (MLP). Once the related features are tailored for artificial neural network training and the hyperparameters are tuned accordingly by deploying an extensive grid search algorithm, the system achieved a good generalization of customer segmentation strategy and hence a good overall accuracy within a few epochs. The proposed system can be integrated into company's data framework such that it can frequently analyse the customer-related data tables and can decide whether a customer is to be promoted or is to remain unchanged. This automatic decision mechanism can improve company's customer satisfaction.