Review Reports - Self-Paced Ensemble-SHAP Approach for the Classification and Interpretation of Crash Severity in Work Zone Areas

Round 1

Reviewer 1 Report

Dear Authors,

In the manuscript, a method based on the SPE and SHAP model to predict the severity of vehicle crash accidents related to the highway work cash was proposed. Overall, the research topic is interesting and valuable. However, there are indeed a few defects, which are attached as follows.

1. The abstract provides too little useful information, and it is suggested to supplement the quantitative description of the advantages of the proposed method over previous methods.

2. There is error in the setting of keywords, and it is recommended to reselect the 4th and 5th keyword.

3. Improper use of Figures. On the one hand, improper size or low clarity of Figures 1, 2, 3, 7, and 10 result in the information contained in these Figures almost unreadable. On the other hand, Figure 6 was set incorrectly and should be correctly set as a Table.

4. On pages 8 and 9, the description and format of the process are inappropriate. It is recommended to use a flowchart.

5. In the manuscript, a large number of results are simply presented in Figures without in-depth and comparative analysis.

6. In the modeling section, there is a lack of necessary explanations on why the SPE and SHAP models were chosen.

Hope my comments may help you. Wish you success!

Sincerely,

The reviewer

Minor editing of English language required.

Author Response

Dear Reviewer,

Thank you very much for reviewing our paper and for taking the time once again to provide thoughtful and insightful comments and suggestions. We have revised the manuscript in response to your comments and suggestions, and we believe that we addressed them to your satisfaction, thus improving the quality and relevance of our paper, please see the attachment.
Best regards,

Author Response File: Author Response.docx

Reviewer 2 Report

Self-Paced Ensemble-SHAP Approach for the Classification and Interpretation of Crash Severity in Work-Zone Related Crashes

I congratulate you for this valuable study on the crash severity of traffic accidents. My reviews on the study are given below.

*I think that the abstract part should be revised and written more simply and plainly in order to make it easier for the reader to understand.

*I think that the contribution and novelty of the study to the literature should be very clearly. It is a fact that the interested readers reach this section without examining the entire article, which has an impact on the reading rate of the article.

*I think that the figural explanations related to the main structure of the study are not sufficiently explained in the text. This section should be improved.

* The limits of the study and the reasons for the limits should be clearly stated.

*I think that the Conclusion and Results sections should be written more emphatically.

Considering the entire article process, it's clear that you've produced a successful job. I congratulate you for this.

Author Response

Dear Reviewer,

Author Response File: Author Response.docx

Reviewer 3 Report

The article deals with a relevant topic, related to the classification and interpretation of crashes, which is relevant to road safety. There are several aspects to improve, so that it can be considered for publication in the journal, which are detailed below:

1. The first recommendation is to reduce the title, there are some words that are repeated like "crash" (crashes).

2. In the abstract, the evaluation of 3 models is initially mentioned, but finally 4 are named. It is understood that there are 4 in total, but it is recommended to improve the wording.

3. The abstract does not identify the novelty of the performed work; it is important to highlight what is relevant to the proposal. It is not appropriate to just mention the algorithms or systems used, it is pertinent to clarify what is new with respect to previous work in this regard.

4. I believe that the final conclusion presented in the abstract should be improved. It should be analyzed much more thoroughly than can be concluded regarding the results obtained, than only presenting which are the less severe types of crashes.

5. In the introduction, the authors present some previous works, regarding the research topic, and at the end they describe procedurally what are the differences of their proposal, but it is not clear what they want to improve, with respect to previous works. This is very important to highlight, as mentioned in point 3, in the abstract in summary, and then in the introduction with much more detail.

6. In section 2, on “materials and methods”, I think it is important to justify why the specific use case is used. Were other possible sources evaluated? It is important to clarify it in the content of the document. In addition, they must justify why the selected ML models were used, how was this selection made?

7. It is recommended to the authors that before section 4 (Conclusions and recommendations) a discussion section be performed, where the relevance of the results obtained is explained, with respect to the research topic.

8. In the conclusions section, it is recommended to eliminate the first paragraphs where the content of the article is summarized again. The conclusions cannot present a summary again, as is done in the abstract.

9. Finally, it is recommended that authors evaluate whether part of the content of sections 2 and 3 can be presented as an Annex, because these sections have too much content.

Regarding the use of English, it is important to improve the writing of several sentences. There are also too long paragraphs, with too many sentences.

Author Response

Dear Reviewer,

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

The authors have adequately met the requested requirements.

A new review is recommended, to identify minor errors.