Abstract: In the past 40 years, demand unconstraining estimation has been continuously developed and sustained attention in the practice of revenue management systems. Considering the business environment in the network organization and the information market environment with the Internet as the distribution channel, the customer booking time and reservation mode have undergone a fundamental change. The existing research results of demand unconstraining estimation methods are summarized. From the perspective of network environment and big data applications, the researches on mathematical statistics, choice-based models as well as machine learning simulation evaluation methods are reviewed and prospected from three aspects: the specific demand distribution form, the strategic customer behavior, and the method actual application robustness and accuracy.

Key words: revenue management system, demand unconstraining estimation, demand distribution, strategic customer behavior, computer simulation

摘要： 近40年间，需求无约束估计在收益管理系统实践中得到了不断发展和持续关注。考虑到在网络型组织业务环境和以互联网为新兴分销渠道的信息市场环境中，顾客预订时间和预订方式发生了根本性转变，总结和梳理了现有需求无约束估计方法的研究成果。以网络环境和大数据应用为视角，从需求所满足的具体分布形式、顾客的策略性选择行为，以及方法实际应用的鲁棒性和准确性三方面，对数理统计类、选择模型类和机器学习仿真评价类方法的研究进行了述评和展望。

关键词: 收益管理系统, 需求无约束估计, 需求分布, 顾客策略行为, 计算机仿真