Enhancing Pattern Classification Accuracy Through Customer Segmentation-Using Machine Learning Algorithms

This research examines and evaluates consumer segmentation clustering algorithms to increase accuracy and comprehensiveness. This study uses multicriteria decision-making and clustering to segment clusters by client purchase behavior. The proposed multi-criteria decision-making (MCDM) approach adds growth rates, consumer collaboration, and the company's desired market strategy to the RFM model's hierarchy of factors. This study preprocesses consumer data using hybrid algorithms like K means++ and affinity propagation. APK-Means++, a hybrid clustering algorithm, combines K-means++ clustering with MRFSC and Affinity Propagation feature selection. Each clustering approach is assessed using the Silhouette Score, Davies–Bouldin Index, and Calinski–Harabasz Index. After comparing Silhouette Scores with K-means++ (0.1674) and Affinity Propagation (0.1237), the hybrid APK-Means++ technique surpasses the standalone methods by 0.3327. The hybrid approach has a lower Davies–Bouldin Index (1.0389) than K-means++ (1.8067) and Affinity Propagation (2.2865), indicating better cluster separation and compactness. The hybrid technique's Calinski–Harabasz Index of 1408.3406 is much higher than Affinity Propagation (156.8100) and K-means++ (566.5759), indicating better cluster quality. These results show how powerful clustering techniques, including hybrid ones, can be for gaining consumer data insights to improve marketing tactics, customer service, and long-term success.