Customer Segment Classification Prediction in the Australian Retail Based on Machine Learning Algorithms

Abstract

Customer segmentation is an essential area of business analytics today. Accurate customer segmentation is access to improves the efficiency of marketing campaigns and customer satisfaction. This study employs multiple machine learning methods to classify Australian Retail company BIGW's customer segments and first to apply multiple model explanation methods to find insights related to customer segmentation identification. After rigorous comparison and hyperparameter fine-tuning, XGBoost is the most adept for this dataset. We derive three key insights through the model results and model interpretive methods. First, BIGW's primary clientele comprises young families in urban areas who prefer cost-effective products, establishing the foundation of their consumer base. Second, the model result indicates a notable gap in BIGW's understanding of its high-end customers, suggesting an area for immediate attention. Third, a specific customer segment emerges from the data: individuals favoring online shopping, demonstrating high total expenditure but low interest in promotions, representing a high-value segment.