服务型产品收益管理中的动态定价研究

发布时间：2018-10-11 19:18

【摘要】：随着经济环境的变动,传统的定价方法在实际运用中出现越来越多的漏洞,无法准确预测消费需求的变化、无法用精准的挖掘潜在消费需求、无法实现企业利润最大化,已经成为服务型企业收益管理的难题。本文主要在收益管理的情景下对服务型产品定价问题进行研究,通过研读和综述服务型产品动态定价的相关文献,明确了服务型产品的特点以及当前服务型产品定价的若干漏洞,并根据当前国内外动态定价的研究成果,归纳了动态定价影响因素并建立本文动态定价模型。之后选取当前热门且符合多维度特性的强化学习方法对模型求解计算,得到本文模型的最优状态动作对,形成环境和行为的最优路径从而得到最大利润。为了验证整体模型和最终结果,本文以服务型产品是否有空间上的移动为标准分情况做了案例仿真和分析,得到不同背景下的最优价格和利润。在固定服务型产品案例上,本文选取酒店客房服务收益管理为背景,建立了酒店客房动态定价模型,该模型结合实际状况将酒店客房剩余数量、消费者策略行为、预售和定价联合考虑进行价格决策,谋求最大化的企业利润。在移动服务型产品的案例上,本文选取高速铁路票价问题作为背景,建立高速铁路客票动态定价模型,模型结合高速铁路站点区间、剩余座位量、消费者策略行为、提前期和价格等多重因素,保证高上座率的同时追求最大化利润。在两个案例寻求最优解的过程中,本文利用强化学习理论和Q学习方法将收益管理的动态定价问题转化成AGENT遍历状态和动作的过程,通过Q学习迭代利润函数得到最优的状态动作对,直接指导企业定价行为。之后将本文的动态定价模型结果与传统定价方法进行比较发现动态定价方法要优于传统定价方法,同时进行了市场、库存、成本、消费者价格敏感程度等因素的敏感性分析。最后对文章进行总结,从动态定价模型、动态定价方法和因素的静态敏感性分析三个方面展开,得到关于企业应利用价格调节作用、关注价格影响因素出现的奇异值等七条结论并提出了相应的管理建议。
[Abstract]:With the change of economic environment, there are more and more loopholes in the practical application of traditional pricing methods, which can not accurately predict the change of consumption demand, can not use accurate mining potential consumption demand, can not achieve the maximum profit of enterprises. It has become a difficult problem in revenue management of service enterprises. In this paper, we study the pricing of service-oriented products in the context of revenue management, and study and summarize the relevant literature on dynamic pricing of service-oriented products. This paper clarifies the characteristics of service products and some loopholes in the pricing of service products at present. According to the current research results of dynamic pricing at home and abroad, the influencing factors of dynamic pricing are summarized and the dynamic pricing model of this paper is established. Then we select the current popular reinforcement learning method which accords with the multi-dimension characteristic to solve the model and get the optimal state action pair of the model to form the optimal path of the environment and behavior so as to obtain the maximum profit. In order to verify the whole model and the final result, this paper makes case simulation and analysis on whether there is space movement of service product, and obtains the optimal price and profit under different background. In the case of fixed service product, this paper selects the income management of hotel room service as the background, and establishes the dynamic pricing model of hotel room. Pre-sale and pricing are considered to make price decisions to maximize corporate profits. In the case of mobile service product, this paper selects the high-speed railway ticket price as the background, establishes the high-speed railway passenger ticket dynamic pricing model, the model combines the high-speed railway station interval, the remaining seat quantity, the consumer strategy behavior. Multiple factors, such as lead time and price, ensure high attendance and maximize profit. In the process of seeking the optimal solution in two cases, this paper uses reinforcement learning theory and Q learning method to transform the dynamic pricing problem of revenue management into the process of AGENT traversal state and action. The optimal state action pair is obtained by Q learning iterative profit function, which directly guides the enterprise pricing behavior. After comparing the results of the dynamic pricing model with the traditional pricing method, it is found that the dynamic pricing method is superior to the traditional pricing method. At the same time, the sensitivity of market, inventory, cost, consumer price sensitivity and other factors are analyzed. Finally, the paper summarizes the dynamic pricing model, the dynamic pricing method and the static sensitivity analysis of factors, and gets the conclusion that enterprises should make use of the role of price regulation. Seven conclusions such as singular value of price influencing factors are concerned and corresponding management suggestions are put forward.
【学位授予单位】：北方工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F224;F274

【参考文献】