Utilizing Ensemble Approach for Predictive Customer Clustering Analysis with Unsupervised Cluster Labeling

Customer clustering is an unsupervised machine-learning approach that groups diverse customers based on shared characteristics. This research focuses on improving customer cluster analysis in the retail sector through an exploration of machine learning techniques, specifically employing the k-means model and predictive algorithms. An Ensemble approach is proposed to gain deeper insights into customer behavior and predict the future actions of new customers within the same cluster assignment. The study evaluates multiple machine learning techniques, utilizing K-fold cross-validation for enhanced model performance, with key metrics including accuracy, precision, recall, and F1-score. Notably, the Extreme Gradient Boosting Classifier excels in Dataset One, while Random Forest outperforms in Dataset Two. The project aims to combine these top-performing ensemble classifiers using the Voting-Soft Classifier for customer classification. The proposed model achieves high precision scores for both datasets, with a particularly promising precision score of 94.57% and an F1-score of 93% for Dataset Two, demonstrating its effectiveness in customer classification.