Abstract: For the scene where the online retailer has multiple warehouses in one region and only considers the case of split due to stock item, is a model established for maximizing the entire order distribution, and the stock item allocation method is studied, to improve the existing algorithms to optimize the items stored in the distribution center to further reduce the order split rate. Aiming at the problem of the Greedy Order Algorithm and the Greedy Hot Sale Algorithm not considering the relationship between items, and combining with Apriori Algorithm, the algorithm is optimized and designed, and a Greedy Association Algorithm is proposed. The algorithm uses a new method of item allocation to find the stock items with strong correlation, and priority is given to the stock items with strong correlation.The experimental results show that compared with the Greedy Order Algorithm and the Greedy Sales Algorithm, the improved algorithm can significantly reduce the rate of order splitting by approximately 8% and 11%, respectively.