Application of Machine Learning Algorithms in Enterprise Supply Chain Demand Forecasting

This study aims to explore the application of machine learning algorithms in enterprise supply chain demand forecasting, focusing on comparing the performance of linear regression, decision tree, and long short-term memory network (LSTM) models. The historical sales data of a large retail enterprise was selected for case analysis. Through data collection and preprocessing, including missing value processing, data standardization, and feature engineering, the above three models were built and trained. The results show that the LSTM model performs well in processing time series data and captures the volatility and trend of sales data. Its prediction results are closer to actual values. In terms of mean square error (MSE), root mean square error (RMSE), and average absolute error (MAE), the LSTM model outperforms decision trees and linear regression models. The research conclusion shows that the LSTM model has significant advantages in demand forecasting tasks and provides an effective tool for enterprises to improve demand forecasting accuracy and optimize supply chain management. In addition, this study also compares the prediction results of different models with actual sales data through data charts, further verifying the superiority of the LSTM model.