MARKETING FRAMEWORK FOR CUSTOMERS CLASSIFICATION

Abstract

Customer classification is a pivotal tool in modern marketing, aimed at optimizing business strategies through consumer segmentation and behavior prediction. The purpose of this study is to analyze the theoretical foundations of customer classification, practical machine learning algorithms, and their application in building predictive models. The paper explores methods such as decision trees (C4.5, ID3), the 1R algorithm, k-nearest neighbors (KNN), Random Forest, and clustering techniques for identifying customer groups with similar characteristics. Using a case study of a travel agency, the research demonstrates how analyzing age and income data enables the identification of optimal advertising materials for different customer segments. The study emphasizes evaluating model quality through metrics (Accuracy, Precision, Recall, F1-score, AUC-ROC) and analyzing confusion matrices, which are critical for minimizing financial losses caused by prediction errors. Marketing strategies (mass, concentrated, differentiated marketing) and their alignment with classification results are discussed in detail. The conclusions highlight the necessity of integrating classification models into decision-making systems to enhance customer loyalty, reduce churn, and maximize revenue. This work holds practical value for marketers, data analysts, and managers seeking to transform data into competitive advantages.