Warehouse Volume Forecasting and Warehouse Segmentation Strategy Based on Time Series and Machine Learning Models

With the booming development of e-commerce, the competition of e-commerce enterprises has intensified, and accurate demand forecasting and reasonable inventory strategy have become the key to improve core competitiveness. In this paper, we focus on e-commerce warehousing, construct ARIMA, decision tree and LSTM models to predict the inventory and daily sales of 350 categories in the next three months, and select the optimal prediction model for each category based on the Mean Absolute Percentage Error (MAPE) to lay a data foundation for warehousing allocation. Then, the target planning models of “one warehouse for one product” and “multiple warehouses for one product” are constructed respectively, and solved by genetic algorithms, taking into full consideration of the warehouse rental cost, warehouse capacity and capacity utilisation rate, category relevance and other factors, to arrive at the corresponding optimal warehousing scheme. These solutions help e-commerce enterprises to optimise the allocation of warehouse resources, reduce operating costs and improve warehouse management efficiency while meeting market demand.