Big Data Approach to Product Demand Prediction Using Machine Learning Based Prediction Models

Abstract

This study explores the application of a machine learning-based big data approach to predict product demand, focusing on a Long Short-Term Memory (LSTM) model developed using big data from the manufacturing and distribution sectors. Improving the accuracy of product demand prediction is the main objective of this study, as prediction accuracy can significantly impact a company's supply chain efficiency and inventory management. Using historical data and real-time data obtained from sales history, search trends, and other external factors, a machine learning model is trained and evaluated using the Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE) metrics. The results show that the LSTM model excels in predicting fluctuating and seasonal demand patterns compared to other algorithms, such as linear regression and decision trees, with an accuracy rate of up to 90% on test data. The big data approach used in this study allows the analysis of external factors that influence product demand, making the prediction model more responsive to market changes. These findings provide significant contributions to supply chain management, reducing the risk of shortages and excess stock, and increasing company profitability. This research is expected to be a reference for companies and academics in developing more adaptive and responsive demand prediction models based on big data.