Overbooking is one type of revenue management techniques that could help reduce costs caused by either undersales or oversales. In this paper, we propose overbooking models for one class of cars and two classes of cars to minimize the total cost for car rentals. In the two classes of cars situation, there is an upgrade possibility for lower class to upper class. This makes the model more complex than one class of cars scenario. We have found that convexity can be proved in both cases. Sensitivity analysis of the parameters is conducted to observe the effects of relevant parameters on the optimal solution. Model simplification is proposed using multiple linear regression analysis, which can help estimate the optimal overbooking level using appropriate independent variables. The results show that the overbooking level from multiple linear regression model is relatively close to the optimal solution (with the adjusted R-squared value of at least 83.40% for model of one class of cars, and 72.8% to 90.10% for model of two classes of car). To evaluate the performance of the proposed model, the total cost was compared with the case where the decision maker uses a na�ve method for the overbooking level. It was found that the total cost from optimal solution only 0.5 to 1 percent (on average) lower than the cost from regression model, while it is approximately 67% lower than the cost obtained by the na�ve method. It indicates that our proposed simplification method using regression analysis can perform very well in estimating the overbooking level.

Username
Password