This research integrates the demand forecasting and inventory control problems to examine how machine learning algorithms enable organizations to elevate supply chain performance. Specifically, we implement the Light Gradient Boosting Machine (LGBM) algorithm to forecast demands characterized by trend and seasonality, comparing its effectiveness against simpler methods. Our findings demonstrate that the LGBM algorithm achieves substantial improvements in forecasting accuracy, as measured by MAE and RMSE, leading to noticeable enhancements in inventory management performance, including considerable reductions in lost sales and excess inventory. We also investigate the impact of the safety stock on these results, revealing that the advantages of advanced forecasting methods become more pronounced as safety stock levels decrease. Understanding the dynamic effects of such advanced forecasting methods allows organizations to make more informed decisions and gain competitive advantages.