A comparative analysis of statistical and machine learning methods for predicting demand in engine oil retailer

Demand prediction plays a crucial role in supply chain management, aiming to enhance accuracy and reliability for optimal operations. This study evaluates various models for forecasting demand in engine oil retail, including traditional statistical methods like Moving Average (MA), Double Moving Average (DMA), Time Series Decomposition (TSD), and Holt-Winters, alongside advanced machine learning techniques such as Artificial Neural Networks (ANN) and Hybrid models. Each model undergoes optimization through techniques like hyperparameter tuning and feature selection. Performance metrics such as Mean Absolute Error (MAE) and Accuracy are utilized to assess their effectiveness in predicting demand patterns.The study focuses on forecasting future demand using ANN and Hybrid model, comparing their performance with traditional methods to determine the most effective approach. The findings highlight the advantages of machine learning over traditional statistical model in capturing complex demand and optimizing supply chain operations.