Tez özetleri Astronomi ve Uzay Bilimleri Anabilim Dalı

Dynamic Pricing For Revenue Management Of Products Which

Have Seasonal Demand In Retailing

Dynamic pricing and revenue management concepts have gained significant importance based on growing appreciation in retailing in recent years. Dynamic pricing problem is described as a retailer’s activity to determine the dynamic prices of a seasonal good in a fixed selling period in order to maximize the revenues while the initial inventory decreases. The change in the customer demand structure is the basic reason for the price shift of a business selling seasonal goods. Such kind of a business should change the price of its seasonal goods in order to maximize its expected revenue. By the way, it is required that the business should analyze the change in the customer demand structure in order to decide how much to shift the prices of seasonal goods. Within this scope, in this thesis, a new methodology which can be efficiently used in the retailing sector have been suggested.

Within the framework of the suggested methodology, for different goods, support vector machine depending on statistical learning with small number of data and poisson regression have been compared in terms of prediction accuracy using mean squared error and tracking signal. According to the result of the comparisons, since the demand functions which have belonged to the better forecasting models have suggested higher revenues, these functions have been used to obtain the price based revenue functions. After this, in the case of no capacity constraints, taking the derivative of these previously obtained price based revenue functions or alternatively using non-constraint nonlinear programming, optimal sales prices have been computed which have maximized the relevant revenue functions. On the other hand, in the case of capacity constraints, the price based revenue functions have been rearranged according to the demand. This arrangement has been made using the relation between price and demand. Then, these rearranged revenue functions have been used as the objective function of the nonlinear programming model and capacity constraints have been added. So, optimal dynamic sales prices which have maximized revenue have been found out. By the way, for the case if the suggested optimal prices may not meet the expected sales, it has been shown how to shift the initial optimal price policy according to the actual sales.
With the above specifications, this thesis brings together and uses techniques of forecasting, statistics, machine learning and operations research for the first time in the literature in the context of dynamic pricing of seasonal goods. As a result, different study areas in industrial engineering have been merged and contributions have been made how to implement dynamic pricing for the retailers selling seasonal goods in Turkey.
In the common sections topic, dynamic pricing and its emergence have been discussed. Then, a wide literature survey has been made and a new methodology which was mentioned above has been suggested to close the gap in the literature. In the third section, the suggested methodology has been implemented in detail. In the fourth section, findings and the results of analysis have been reported and in the last section, experiences and contribitions have been explained.
December 2013, 117 pages

Keywords: Dynamic pricing, revenue management, optimization