销量预测作为企业营销决策和战略管理中的重要环节,一直是研究的热点。但由于销量受企业内部环境和外部环境中多种因素影响,销量预测一直是极具挑战性的研究问题。现有的销量预测模型常使用商品的历史销量和商品属性等变量来预测销量,很少考虑利用其他商品信息来提高预测精确度。本文首先通过市场购物篮分析,找出销量相互影响的商品品类。然后根据市场购物篮分析结果构建品类之间的关联网络,并利用网络中与待预测品类相关联的其他品类的销量信息来提高预测准确度。为了解决顶测过程中存在的内生性问题,本文采用向量自回归模型对预测问题进行建模,同时控制节假日等因素对销量的影响。本文用一家国内大型超市的真实数据进行验证,结果表明本文提出的方法比传统方法具有更高的精确度。最后,本文基于得到的研究结果,为企业的库存管理和营销策略提出一些管理建议。 As an important part of marketing decision-making and strategic management, sales forecasting has always been a hot spot for research. However, sales volume has been a challenging research issue due to the impact of sales volume on many factors in the enterprise’s internal and external environment. Existing sales forecasting models often use variables such as the historical sales of goods and the attributes of goods to predict sales, seldom consider using other goods information to improve forecasting accuracy. This article first through the market basket analysis to identify the interaction between sales of product categories. Then construct the network of links between categories based on market basket analysis and use the sales information of other categories in the network associated with the category to be predicted to improve the prediction accuracy. In order to solve the endogeneity problem in the top test, we use vector autoregressive model to model the forecast problem and control the influence of the factors such as holidays on the sales volume. In this paper, the real data of a large domestic supermarket is used to verify the results. The results show that the proposed method has higher accuracy than the traditional method. Finally, based on the research results obtained, this paper puts forward some management suggestions for the enterprise’s inventory management and marketing strategy.