In industrial retail management, poor demand forecasting and ineffective inventory control result in loss of revenue and wasted resources. This study formulates a machine learning-based consumer behavior forecasting model to facilitate real-time decision-making, maximize stock levels, enhance supply chain efficiency, and improve customer satisfaction through personalized suggestions. Data is gathered from various sources such as Point of Sale (POS) systems, web analytics, customer buying history, supply chain databases, and IoT sensors to collect transactional, behavioral, and inventory data. The dataset is preprocessed, where missing values are addressed, outliers are eliminated, and feature selection is performed to improve model performance. Four machine learning algorithms Random Forest (RF), Artificial Neural Networks (ANN), Support Vector Machine (SVM), and K-Means Clustering are trained and tested based on accuracy, precision, recall, F1-score, Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE). The accuracy shows that K-Means Clustering has the best accuracy of 97.6%, followed by SVM at 93.4%, RF at 89.7%, and ANN at 84.5%. K-Means also has the lowest MAE (0.024) and RMSE (0.037), proving best predictive performance. The confusion matrix also confirms its effectiveness in minimizing false predictions. The best model is incorporated into a decision support system to facilitate real-time demand forecasting, inventory optimization, and personalized customer interaction. This research points out the critical role played by machine learning in revolutionizing retail management through enhanced operational efficiency and data-driven decision-making.