系统管理学报

系统管理学报 ›› 2023, Vol. 32 ›› Issue (1): 23-41.DOI: 10.3969/j.issn.1005-2542.2023.01.002

• 决策科学与运营管理 • 上一篇    下一篇

基于顾客到达和购买数据的新品广告预算分配学习算法

高秋爽，黄帝媛，杨超林   

  1. 上海财经大学信息管理与工程学院 交叉科学研究院，上海 200433
  • 出版日期:2023-01-28 发布日期:2023-01-17
  • 作者简介:高秋爽（1998-），女，硕士。研究方向为供应链管理。
  • 基金资助:
    国家自然科学基金资助项目（72122012，72071126）

Learning Algorithm for New Product Advertising Budget Allocation Based on Customer Arrival and Purchase Data

GAO Qiushuang，HUANG Diyuan，YANG Chaolin   

  1. Research Institute for Interdisciplinary Sciences, School of Information Management and Engineering, Shanghai University of Finance and Economics, Shanghai 200433, China
  • Online:2023-01-28 Published:2023-01-17

PDF

摘要：

新产品上市导入期，企业通常会设定一定预算进行广告营销。该时期产品的销售过程可划分为“获客”和“转化”两个阶段，广告首先将潜在消费者引流到消费平台上，消费者再根据产品特性以一定概率达成购买行为。由于欠缺历史数据，企业难以评估消费者对新产品广告投入的反应以及产品本身对消费者的吸引力。为解决给定广告总预算和总库存约束的新产品推广期广告预算分配的问题，提出了一种利用顾客到达和购买数据，同时学习顾客到达人数与广告投入的关系以及顾客购买概率（转化率）的非参数学习算法。从理论上证明了预算分配策略的渐进最优性质，并通过数值实验验证了策略在多种场景下的性能，说明了策略的鲁棒性。

关键词:

广告预算分配, 非参数学习算法, 渐进最优策略

Abstract:

During the introduction period of a new product, companies usually set a certain budget for advertising. The product sales process in this period can be divided into two stages: “customer acquisition” and “conversion”. Advertising first attracts potential consumers to the consumption platform, and then consumers purchase with a certain probability depending on the product characteristics. To solve the advertising budget allocation problem for the promotion of a new product given the total advertising budget and total inventory constraint, this paper proposes a non-parametric learning algorithm that uses customer arrival and purchase data to simultaneously learn the relationship between customer arrivals and advertising, as well as the probability of product purchase. It is proved that the learning algorithm proposed is asymptotically optimal. Numerical experiments verify its performance in a variety of scenarios, illustrating the robustness of the strategy.

Key words:

advertising budget allocation, nonparametric learning, asymptotic optimality policy

中图分类号: