An Application of a Deep Q-Network Based Dynamic Fare Bidding System to Improve the Use of Taxi Services during Off-Peak Hours in Seoul
Round 1
Reviewer 1 Report
This is an extremely high-quality article that addresses using the DQN-based Dynamic Fare Bidding System to improve the Taxi Service during off-peak hours in Seoul, Korea. The problem is considered both on a theoretical level and tested on an existing model with the help of simulation.
It is evident from the article that the problem is considered more from the point of view of an adaptation of operators or the point of view of an adaptation of the Fare Bidding system in the considered environment. Furthermore, it is not possible to understand in what way it is possible to evaluate the acceptance of users for changes in the service. In my opinion, service users are the ones who can make a significant contribution (decide) to the success of the proposed system with their willingness to pay. I suggest a short supplement to the article in this direction.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Dear Authors,
The scientific goal of increasing the demand for use and further increasing drivers' revenues through real-time tariff discounts for each regional time zone during the off-peak daytime in Seoul has been achieved. The presented manuscript describes the issues of modeling the availability of taxi vehicles quite well, due to the operational parameters of the vehicles. The methodological assumptions and simulations were presented by the authors in a very good and proper way. Much attention was paid to the choice of tariffs and benefits for drivers in the presented districts.
The methodological assumptions and simulations were presented by the authors in a very good and proper way. Much attention was paid to the choice of tariffs and benefits for drivers in the presented districts. Unfortunately, I lacked information on the number of vehicles tested in relation to the total number of vehicles operating in this area. What is the value of the sample in relation to the total number of registered carriers (taxi). Is this market highly regulated or are there also drivers operating without a license?
Please state why 50,000 episodes were selected for the simulation.
This information is important for the suitability of the simulation for mapping in real traffic conditions. Please complete this information in the methodology section.
In addition, there are some editing errors (no spaces between the text and the references in square brackets). Please review the text under this account.
Thank you for getting to know your work.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Dear authors,
Please find in the attached file my comments to this manuscript.
Thank you.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Dear authors,
Please find my comments to this revised version in the attached file.
Thank you.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 3
Reviewer 3 Report
Dear authors,
Please find my comments to this new version in the attached file.
Thank you.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Thanks to the reviewer, the quality of our paper seems to have improved greatly. Thank you for pointing out the deficiencies and suggesting specific improvements.
Author Response File: Author Response.pdf
