Short-Term Traffic Flow Prediction Based on CNN-BILSTM with Multicomponent Information
Round 1
Reviewer 1 Report
This study proposes a multistep prediction model based on a convolutional neural network and bidirectional long short-term memory (BILSTM) model. The spatial characteristics of the data were considered as input of the BILSTM model to extract the time-series characteristics of the traffic. The experimental results validated that the BILSTM model improved the prediction accuracy in comparison to the support vector regression and gated recurring unit models. Overall, the manuscript is well written and interesting; however, the following issues should be addressed before the final publication.
1. For Section 1, the authors should provide the comments of the cited papers after introducing each relevant work. What readers require is, by convinced literature review, to understand the clear thinking/consideration of why the proposed approach can reach more convincing results. This is the very contribution from the authors.
2. There are some typos and grammatical mistakes in the manuscript, please correct them.
3. Please provide the section-wise breakup at the end of section 1.
4. The equations are not properly written and should be corrected.
5. Why did the authors consider only ReLu function as an activation layer function?
6. The heading of section 2 should be Methods.
7. The caption/title of the figures is missing.
8. How are the hyper-parameters of the models selected in Section 4?
9. Why is the data only considered from May 1 to July 31, 2021? Your model performance on other months may be very different.
10. What was the testing period? How many days? Similarly, how many days are used as a training data set? Did you use the rolling window technique? Please clarify all this in the manuscript.
11. Correctly number the tables.
12. Looking at the MAPE in the table, It is quite interesting the model produced better results for 30 minutes prediction intervals than for 15 minutes. In general, these kinds of results are not common. Can you explain this?
13. In the literature review, it is highly recommended to discuss other time series models and methods to highlight the importance of this topic for the reader of this journal. Some recommendations that can be considered are (10.1155/2022/6709779) (10.3390/en15093423)
Author Response
We would like to thank you for your careful reading, helpful comments, and constructive suggestions, which has significantly improved the presentation of our manuscript.
We have carefully considered all comments from the reviewers and revised our manuscript accordingly. The manuscript has also been double-checked, and the typos and grammar errors we found have been corrected by the professional organization Editage. In the following section, we summarize our responses to each comment from the reviewers. We believe that our responses have well addressed all concerns from the reviewers. We hope our revised manuscript can be accepted for publication.
Great thanks for your comments.
Author Response File: 
Author Response.docx
Reviewer 2 Report
1. There are no figure captions in this paper.2. The authors introduced various papers in Sec. 2. However, the reviewer thinks that discussion for the papers is insufficient because the authors give small explanation about them. In addition, there are no information that the existing works for the traffic prediction are insufficient or inappropriate for the target field. The reviewer thinks that the comparison is conducted only using general algorithms which do not focus on the specific application such as traffic flow prediction. The reviewer suggests two options to improve this paper as below.
2-1) If the existing works are inappropriate or not be able to apply them to the target field, please add the explanations to this paper.
2-2) If the existing works are insufficient, please compare with them or please explain enough reasons why they cannot achieve the better performance than the proposed algorithm.
Author Response
We would like to thank you for your careful reading, helpful comments, and constructive suggestions, which has significantly improved the presentation of our manuscript.
We have carefully considered all comments from the reviewers and revised our manuscript accordingly. The manuscript has also been double-checked, and the typos and grammar errors we found have been corrected by the professional organization Editage. In the following section, we summarize our responses to each comment from the reviewers. We believe that our responses have well addressed all concerns from the reviewers. We hope our revised manuscript can be accepted for publication.
Great thanks for your comments.
Author Response File: Author Response.docx
Reviewer 3 Report
The development of an intelligent model of a city's transport system is an essential element of spatial planning in urban areas. Available models of vehicle flows are not sufficient with rapidly changing information related to the movement of many transport modes.
Therefore, the authors of the reviewed article proposed a multi-stage prediction model based on a conventional neural network (CNN) and a bidirectional short-term memory (BILSTM). To validate this model, they conducted real-world tests and compared the obtained prediction results of the CNN-BILSTM model with the other three traffic models.
The article's authors have presented what they have realized in a clear and readable manner. They have reviewed the literature in an appropriate manner. The conclusions are appropriate. I, therefore, believe that a peer-reviewed article should be published in Applied Sciences.
Author Response
We would like to thank you for your careful reading, helpful comments, and constructive suggestions, which has significantly improved the presentation of our manuscript.
Great thanks for your comments.
Round 2
Reviewer 1 Report
I have no further comments. Thank you
