Outsmarting Human Design in Airline Revenue Management
Round 1
Reviewer 1 Report
This paper investigates the application of RL to maximize the revenue of aircraft companies by properly fixing the prices of the tickets. Although this problem has been widely analyzed in the literature, the approach of the authors presents a clear novelty for this important problem.
However, the manuscript needs to address some issues to finally consider the paper for publication:
1- The manuscript should be presented in a more formal or academical manner. Sentences like:
Page 3 Line 100 "It's not only..." should be written "It is not only"
Page 11 Lines 406-407 (obviously, the computation). This kind of writing is not appropriate for scientific writing.
Also, there is not a clear definition of which formulas are included in the main text and which ones are part of the text which is confusing for the reader.
Please revise the manuscript with a formal writing sight.
2- On Page 5 Lines 200-202: "For convenience, if the flight's capacity is exhausted before departure, interactions continue, however no further bookings are possible" Please explain why interactions continue for convenience.
3- On Page 5 Line 207: Authors declare that for the RM, the discount rate is usually set to 1. Please demonstrate in the literature this claim. As the authors later claim, this value is not possible for an infinite horizon. Please discuss this claim in Section 3.1.
4- In general, the values of the hyperparameters of the algorithm implemented are not presented. Please provide this values to help the reproducibility of the experiments.
5- On Page 8 Lines 301-302: "Among the many RL methods in literature, the actor-critic methods are well suited for large action spaces". Please provide references for validating this sentence. In addition, a more detailed justification of the implementation of this algorithm for the single-leg problem should be given.
6- In Section 5, frat5 is introduced and there is not a clear reference to explaining the meaning of this term.
7- In Figure 7, following the MDPI guidelines, the authors should also define the term MSE and in the main text later also doing this definition.
8- An study analyzing the evolution of the prices over a particular flight along the time horizon, analyzing the remaining capacity, for the heuristic and RL methodologies could help interpret the different decisions of both algorithms.
9- An analysis of the interactions among the decisions of the RL algorithm while fixing the prices for all the available flights could be really interesting to interpret the combined solution of the RM problem. For instance, while other flights are completed in capacity, the algorithm is risker with the remaining capacity flight since revenue has been guaranteed.
10- Also, the information of the software to take the decision of booking a flight could be better introduced in the manuscript.
Overall, as stated, it is a very good paper and well-structured. However, I encourage the authors to address these comments to improve the overall quality of the manuscript and consider it for final publication in the journal.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
The study is good.
- Provide at least 5 references from 2022.
- Provide a contribution table by comparing this study with the literature.
- In Section 5, provide scenarios in subsections/ subsubsections for understanding.
- Graphical representations are good. Besides, summerise results in a table also.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Dear authors,
I appreciate the novelty of your research and the effort for improving the quality of your manuscript in the revised version.
Consequently, I do recommend the publication of the paper in its current form.