Enhancing Supply Chain Management in Manufacturing Plants: Anomaly Detection and Mitigation Using MCDM and Machine Learning Techniques

Abstract

Supply chain (SC) anomalies, such as inaccurate demand forecasts, inventory imbalances, and production delays, impede operational efficiency and financial performance in manufacturing plants, particularly in third-world contexts where data-driven forecasting methods remain underutilized. This study investigates SC anomalies within Fair Electronics, a key manufacturing partner and authorized distributor of Samsung products in Bangladesh. Through interviews with technicians, supervisors, and management, specific anomalies such as demand volatility, stockouts, and inefficiencies in resource allocation were identified, with a predominant issue being the reliance on experiential forecasting methods that often result in inaccurate demand predictions. Leveraging insights from an extensive literature review, this research introduces machine learning (ML)-based forecasting methodologies tailored to these challenges. Four ML models, including an autoregressive integrated moving average (ARIMA), extreme gradient boosting (XGBoost), long short-term memory (LSTM), and Prophet, were applied to diverse market segments of mobile products, with model evaluation based on metrics such as mean absolute error (MAE), mean squared error (MSE), root mean squared error (RMSE), and mean absolute percentage error (MAPE). XGBoost consistently emerged as the superior model in terms of forecasting accuracy and robustness. The study highlights the transformative potential of advanced ML techniques in enhancing demand forecasting within SCs, proposing a comprehensive framework that integrates these methods to optimize inventory management, production planning, and overall operational performance. This study bridges the gap between traditional and data-driven forecasting approaches, providing a robust evidence base for the adoption of ML in SCs operations, paving the way for enhanced decision-making, reduced inefficiencies, and improved financial outcomes in manufacturing environments similar to Fair Electronics. The findings also offer a roadmap for future research and practical applications in the evolving landscape of supply chain management (SCM).