Dynamic NeuroFusion Ensemble Prediction System for Retail Sales and Customer Behaviour Prediction
ID：76
View protection：Participant Only
Updated time：2026-07-22 16:09:42 Views：11
In-person
Abstract
Abstract—The increasing complexity of retail operations and evolving market patterns has created a significant need for intelligent prediction models that support data-driven decision-making. However, conventional Machine Learning (ML) and Deep Learning (DL) models often encounter difficulties in handling heterogeneous retail data, nonlinear feature relationships, redundant variables and complex predictive patterns. This research proposes a Dynamic NeuroFusion Ensemble Prediction System (DNEPS) for retail product-demand classification. DNEPS incorporates a NeuroWeave Feature Intelligence Layer (NFIL) for adaptive feature-interaction learning, an Adaptive Performance-Guided Ensemble Layer (APGEL) for assigning validation-based weights to individual learners and an Intelligent Fusion and Decision Layer for generating integrated predictive insights. The framework combines Random Forest, XGBoost, LightGBM and Deep Neural Network (DNN) models within unified ensemble architecture. The proposed system was evaluated using a large-scale synthetic Retail Sales and Customer Behaviour Analysis dataset, with product demand categorized into low, medium and high-demand classes. Leakage-free preprocessing and training-derived percentile thresholds were used during model development. Experimental results indicate that DNEPS achieves 98.6% accuracy, outperforming the selected baseline models. These findings demonstrate the effectiveness of adaptive feature intelligence and performance-guided ensemble fusion for reliable retail demand classification and decision support.
Keywords
Keywords—Adaptive Ensemble Learning, Decision Intelligence, NeuroWeave Feature Intelligence, Organizational Forecasting, Predictive Analytics, Retail Analytics
Post comments