Improved Long-Term Forecasting of Passenger Flow at Rail Transit Stations Based on an Artificial Neural Network

Round 1

Reviewer 1 Report

The paper titled "Improved long-term forecasting of passenger flow at rail transit stations based on ANN" by Du et al. aims to address the challenges associated with long-term passenger flow prediction in rail transit stations, which is crucial for planning and infrastructure development. 

However, I have the following suggestions:

1. What is OD? The authors should write down the full abbreviation for the first time you use it, please check.

2. What are the features and the dataset you mentioned? Will it be publicly available? I suggest that you make a table to describe above since it's very critical.

3. Why ANN? Do you try something like LSTM or GRU?

4. It seems to me that lot of discussion should be added. e.g. which line has the best prediction result and why?

5. Can it be applied into any city? How and why?

6. It seems to me that the writing skills should be improve to interest readers.

N/A

Author Response

Greetings esteemed reviewer:
Thank you very much for your suggestions on my paper, I have now sent you the draft response as well as the revised version of the paper as two attachments. I hope you can give me your substantive comments when you are not busy.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Authors,

The paper is well structured, coherent and logical.

It contains numerous figures, which significantly increases the cognitive value of the work.

The research methods are correctly applied.

In general, the paper provides an interesting point of view of improved long-term forecasting of passenger flow at rail transit stations, however it needs minor improvement.

1. There is no clear goal of the research in the abstract. In fact this goal is at the end of this article line 377-381.

2. The authors should more emphasise the research gap.

3. On what theory do the authors base their considerations?

4. Why don't the authors provide any hypotheses?

5. I strongly suggest to divide the section 4: Discussion and Conclusions

6. In the "Discussion part", the authors should compare and contrast the similarity/dissimilarity of findings in this research to the findings of other studies. This way, the authors can argue how this study stands among related studies, convincing readers of the originality or confirmatory values of this study toward the literature.

7. In the conclusions part the authors should confirm the achievement of the research goal.

Author Response

Greetings esteemed reviewer:
Thank you very much for your suggestions on my paper, I have now sent you the draft response as well as the revised version of the paper as two attachments. I hope you can give me your substantive comments when you are not busy.

Author Response File: Author Response.pdf

Reviewer 3 Report

Title: The title of the paper clearly defines the subject of the paper.

Abstract: In the abstract, the research objective, potential problem, method, and research results applied in the paper are clearly indicated. However, the authors did not mention the most significant influencing factors on passenger flow obtained by the model.

 Introduction: Through the introductory part of the paper, the authors made it clear how important it is to predict the flow of passengers during the construction of railway stations. The authors made an effort to comprehensively, through a review of the literature, indicate the prevalence of the considered topic by many researchers.

 Analysis of influencing factors: The authors paid too much attention to the description of influencing factors. In a very detailed and clear way, the authors gave arguments regarding the factors influencing the possible flow of passengers.

 Research Methods and Model Construction: The authors did not pay enough attention to the description and clarification of the application of artificial neural network models. One gets the impression of a vague description of the structure itself, that is, the architecture of the network and its components, such as neurons, layers, activation functions, and the like. Why did you take 80% of the training database? What is the sample considered in this paper? How many layers does your multilayer perceptron have? Did you define the number of hidden layers yourself or did you leave that role to the software? Have you identified the database intended for validation with the data for testing? It seems that the model itself is not the most clearly explained, i.e. there is a lot of information missing that would help the reader draw a conclusion about the created ANN model. In addition to the lack of a Results chapter, the authors present the obtained results of not only the neural network model but also the results of the applied two CART trees and the Chaid tree, which were not even mentioned before. The difference between the mentioned models, i.e. the way they function and the reason for their comparison, is not defined. Also, when presenting the importance of influential factors, there are missing normalized values whose percentages would specify the importance of the obtained results.

Discussion and Conclusions: The discussion of the obtained results should compare in more detail the obtained results and the differences between the applied models, which is not the case here. Also, the shortcoming of this part of the work refers to the comparison of the results with similar studies. This part of the work should be separate. In the conclusion, the authors put a greater focus on the analysis of influential factors, in relation to the prediction of passengers.

Author Response

Greetings esteemed reviewer:
Thank you very much for your suggestions on my paper, I have now sent you the draft response as well as the revised version of the paper as two attachments. I hope you can give me your substantive comments when you are not busy.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have addressed the majority of the concerns, and I concur with the decision to accept the paper.

Author Response

Thank you very much for accepting my paper. Your encouragement and suggestions have helped me a lot!

Reviewer 3 Report

I'm glad you considered the suggestions. However, although the two models CART tree model and the Chaid tree are mentioned in the paper only for comparison, it will not be clear to the readers how the results were obtained, that is, how the models function. With that, I ask you to supplement your work like the ANN model with a description of the application of the remaining two. Also, although you have separated the discussion as a separate one, I suggest that you compare your results with the results of other studies that applied the same method, that dealt with the same topic. This would have a significant impact on the improvement of work.

Author Response

Dear esteemed reviewers:
Thank you very much for your suggestion. In response to your suggestions, we have made targeted changes. The attached document contains a targeted response. In addition, we have marked yellow and resubmitted the modified part of the paper.

Author Response File: Author Response.pdf