Insufficient accuracy of demand forecasting in current enterprise supply chain management leads to inefficient inventory management and high cost. In this study, we constructed an enterprise supply chain demand forecasting model based on the random forest algorithm and designed an inventory cost control system combined with the particle swarm optimization algorithm to solve the cost control problem caused by inaccurate forecasting in enterprise inventory management. The optimal feature subsets are screened by two feature selection algorithms, MCMR and rMCMR, and Random Forest is used to forecast the demand for three types of FMCG products in Company K. PSO-RF is used to optimize the inventory cost control. The results show that the random forest prediction model has the best prediction accuracy of 93.3% in the comparison of multiple classification algorithms, and the training time is only 26.148 seconds; in terms of inventory cost control, the fulfillment cost rate of Company K continues to decline after the adoption of big data technology, and reaches a historical low of 6.286% in 2021, with a significant increase in inventory turnover rate. Company K’s user satisfaction score reaches 4.548, with a platform feedback rate of 100%, a comprehensive rating of over 0.9, and a “recommended order” rating. The study proves that Random Forest algorithm combined with PSO optimization can effectively improve the accuracy of enterprise supply chain demand forecasting, optimize inventory cost control, and enhance the operational efficiency and user satisfaction.
