A Unified Intelligent Analytics Platform for Enterprise Retail Monitoring, Evaluation, and Insight Generation

Abstract

The rapid growth of corporate retail systems has resulted in the generation of large volumes of transactional, demographic, and customer feedback data, making efficient analysis essential for business success. Understanding customer behaviour, improving product performance, and enhancing sales outcomes require intelligent data-driven solutions. This research focuses on developing an advanced retail analytics framework capable of predicting customer ratings and total sales while supporting effective decision-making. Traditional retail analysis relied on manual approaches such as spreadsheets, static reports, and basic Business Intelligence (BI) tools. These methods often failed to capture insights from customer reviews and depended on historical averages or expert assumptions for forecasting. Consequently, they were time-consuming, less accurate, and lacked real-time predictive capabilities, resulting in delayed and reactive business strategies. To overcome these challenges, the proposed system introduces a Machine Learning (ML) based end-to-end analytics platform. Retail data is processed through automated pipelines that manage numerical, categorical, and textual features efficiently. Customer reviews are analysed using Term Frequency–Inverse Document Frequency (TFIDF) to enable sentiment-aware predictions. Advanced models such as Hierarchical Shrinkage (HS) Tree, Tree Alternating Optimization (TAO) Tree, and Boosted Rules Classifier (BRC) are utilized for both classification and regression tasks. Additionally, a hybrid model combining Graph Convolutional Network (GCN) with Natural Gradient Boosting (NGB) is proposed to capture complex feature relationships. Experimental results demonstrate improved performance in terms of Root Mean Square Error (RMSE) and R². The system is implemented using Flask, integrating Exploratory Data Analysis (EDA), model training, and real-time prediction into a unified platform.