Masters Theses


"This thesis presents a novel semi-Markov actor-critic algorithm to solve the airline revenue management (ARM) problem. The ARM problem has been the subject of much research since the deregulation of the airline industry in 1978. Various heuristics and numeric techniques have been developed and continue to be developed for solving this problem, but the search for more efficient techniques is still ongoing. This work focuses on the single-leg ARM problem modeled as a semi-Markov decision problem (SMDP). For Markov decision problems that suffer from the curses of dimensionality and modeling, it is difficult to obtain solutions with classical dynamic programming. In many practical problems, the time spent in each transition of the underlying Markov chains is itself a random variable, which makes the problems SMDPs. To solve the ARM problem under the long-run average reward criterion, this thesis employs the actor-critic algorithm, which is a well-known reinforcement learning algorithm. This is the first attempt at using an actor-critic algorithm for solving an average reward SMDP. The results obtained show that the new algorithm clearly outperforms the EMSR-b heuristic, which is a method widely used in the airline industry. These findings are significant for both revenue management and reinforcement learning techniques. The methodology used here is applicable not only to the airline industry, but also to revenue management in the car rental, hotel, and cruise line industries"--Abstract, page iii.


Gosavi, Abhijit

Committee Member(s)

Qin, Ruwen
Grantham Lough, Katie, 1979-


Engineering Management and Systems Engineering

Degree Name

M.S. in Engineering Management


Missouri University of Science and Technology

Publication Date

Spring 2011


viii, 51 pages

Note about bibliography

Includes bibliographical references (pages 48-50).

Geographic Coverage

United States


© 2011 Ketaki Dilip Kulkarni, All rights reserved.

Document Type

Thesis - Open Access

File Type




Subject Headings

Airlines -- Cost of operation -- United States
Airlines -- Management -- Mathematical models
Markov processes -- Numerical solutions

Thesis Number

T 9829

Print OCLC #


Electronic OCLC #


Link to Catalog Record

Electronic access to the full-text of this document is restricted to Missouri S&T users. Otherwise, request this publication directly from Missouri S&T Library or contact your local library.