Short-Term Traffic Flow Forecasting Method Based on Secondary Decomposition and Conventional Neural Network–Transformer

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors This manuscript proposes a novel short-term traffic flow forecasting method based on secondary decomposition and CNN-Transformer model. The proposed hybrid model is mainly composed of CEEMDAN-VMD and CNN-Transformer. The idea of the paper is clear and innovative and the paper is technically correct, but there are some issues to be improved:
(1) The characteristics of CEEMDAN and VMD algorithms should be explained in the paper.
(2) What do NBXX16 (2) and NBXX11 (3) represent？
(3) Revise the typesetting of the mathematical formulas. For example, in lines 222-223. Formulas need to be displayed in the center and the formula number should be displayed on the right side.
(4) Overall, the English language is fine, but some grammar errors need to be corrected. For example, in lines 279-281“the Transformer model identifies relationships within time series data more effectively and quickly compared to RNN-based models, extracting a wealth of feature information from the data.” In line 364, “Short-term traffic flow data” should be changed to “short-term traffic flow data”.
(5) “Rolling multistep forecasting” in Figure 16 should be changed to “Iterative multistep forecasting”. Comments on the Quality of English Language

none

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript is written consistently and logically. The structure of the article is correct. Based on a broad literature review, the authors identified a research gap and determined their contribution in the field of Short-Term Traffic Flow Forecasting.

The substantive methodological part was well described. The results were presented in an interesting way and commented appropriately.

I suggest that section 3.7 be transformed into section 4. Discussion.

It is also worth indicating directions for further research.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

 

The topic is interesting, the work structure and the scientific content of the paper is good, the academic level of the paper is good, the conclusions are justified. Overall, this paper is of a good quality and well-written. The work can be potentially considered after all the above-mentioned questions are properly addressed. However, the current paper is not conditionally publishable, and should be further revised.

1. Is the optimization result verified with the experimental result?

2. How to ensure that results obtained by ANN model proposed by the authors are reliable?

3. For an artificial neural network model, three main data sets are required: training set, validation set, and test set. How data is selected and what is the basis for selection？

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

Comments and Suggestions for Authors

The authors present the article entitled “Short-Term Traffic Flow Forecasting Method Based on Secondary Decomposition and CNN-Transformer”

This paper presents a short-term traffic flow forecasting method based on secondary decomposition and CNN-Transformer is proposed. 

The article presents the following concerns:

• Add hyperlinks to tables, figures, and references.

• Abstract section: I suggest adding quantitative values in order to highlight the main results.

• Introduction section must improve.I suggest that this section be synthesized so that it can show a broad context. In addition, consider it appropriate that the controversy could include quantitative data. For example, how relatively small is this data?. At the end of the Introduction, the authors can place the main contributions, the scope of the work and the structure of the manuscript.

• Please present the literature review in a single section. Also, provide quantitative results of the reviewed works in order to present a deep analysis of the state-of-the-art and highlight the main findings.

• Figure 2: The CNN configuration is similar to the proposal at https://doi.org/10.1016/j.energy.2019.116316. It is recommended that you justify the use of this configuration in the proposed work and also cite the original source.

• You have written the same word “nonlinear” with and without a hyphen in your document. Both ways are acceptable, but it’s best to be consistent.

• The word “Table #” should be capitalized in this text.

• Please mention future works in the conclusion section.

• In general, the originality of the article is not clear. The manuscript presents 47% of duplicity level. For this reason, I can not extend my recomendation for publication.

Comments on the Quality of English Language

The following misspellings should be checked:

1. line 81: “It is difficult…” may sound overly negative to your reader. Consider rephrasing it as “it isn’t easy..”

2. line 111: “Chen at al. [23] proposed…” The word “at” doesn’t seem to fit this context. Consider replacing it with a different one: “Chen et al. [23] proposed…”

3. line 124: the word “difficult” is often overused. Consider using a more specific synonym to improve the sharpness of your writing: “challenging”

4. line 256: “The channels of the convolutional layers are set to 4, 8 and 16 respectively…” This sentence contains a few punctuation mistakes. Consider changing by “The channels of the convolutional layers are set to 4, 8, and 16, respectively…”

5. line 312: “The structure can be roughly described as in Figure 3…” should be rewritten by “The structure can be described in Figure 3…”

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 4 Report

Comments and Suggestions for Authors

The authors provide an updated version of the manuscript. My previous concerns were addressed. They introduce a CNN-transform method and subsequently validate its performance using cross-validation. The similarity level decreases and improves in quality. However, here are some concerns:

1. Lines 36-37: I suggest avoiding colloquial expressions like 'isn’t easy work', in order to maintain formality and precision in the manuscript.

2. Line 81: “decom-position” should be “decomposition”.

3. Line 359: Please provide a deeper description for Figures 7, 8 and 9 separately.

4. I suggest adding a table that compares the main contributions of the proposed work vs the already reported state-of-the-art in order to highlight the novelty of the work.

Figure 6 is the same figure from  https://doi.org/10.1155/2018/3093596. Please see https://www.mdpi.com/authors/rights for Reproducing Published Material from other Publishers. I suggest the authors create or redraw with significantly changes their own Figure or request permission for using artworks.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 3

Reviewer 4 Report

Comments and Suggestions for Authors

The authors addressed all my concerns, thank you. The manuscript is ready for publication.