Segmenting customers and predicting loyalty has become a powerful way for companies to work on customer retention and marketing optimization. This study goes through a hybrid model, which is a combination of clustering, fuzzy logic, and machine learning, in calculating customer loyalty scores with a focus on their behavior and sustainability in virtual worlds. Customer data are therefore preprocessed in terms of methodology: categorical feature encoding, numerical variable normalization, and treatment of missing values. In the customer segmentation phase, K-Means clustering is performed using purchasing patterns and behavioral characteristics. Next, the fuzzy logic system scores loyalty dynamically based on the behavior incidence and features specific to the segment, thus providing flexibility and interpretability over core threshold-based methods. To predict loyalty scores, the Random Forest Regressor is trained with features including purchase amount, purchase frequency, customer reviews, and transaction history. The model validation includes visual analysis of the cluster distributions, ranking of feature importance, and histograms of loyalty scores. The results prove that the combined methods of clustering, fuzzy logic, and machine learning enhance the accuracy and interpretability in customer segmentation and loyalty prediction. Data-driven methods, in this case, find a great applicability in intelligent consumer analytics with regard to virtual set-ups, where consumer behavior and sustainability are of immense significance. Businesses can leverage these findings to craft targeted marketing campaigns, stimulate customer dedication, and encourage habits of sustainable consumption.