Time Series Analysis of Product Demand Forecasting and Inventory Optimization on E-commerce Platforms

With the continuous advancement of reform and opening up, China's economy has welcomed rapid development, and e-commerce platforms have sprung up like bamboo shoots after rain. The purpose of this study is to use time series models to forecast demand and optimize inventory for thousands of merchants, goods, and supporting warehouses on the e-commerce platform. First, an ARIMA time series model is established for the shipment of old products over time, and through continuous iteration, the optimal parameters of the time series model are obtained for predicting the old products. Then, using K-means clustering, the final prediction results are categorized. Later, new products replace the old ones, and after extracting the feature values of both new and old products to conduct cosine similarity analysis, adjustments are made to the new prediction model to obtain the final forecast values.