Short-Term Traffic Flow Prediction Based on a K-Nearest Neighbor and Bidirectional Long Short-Term Memory Model
Round 1
Reviewer 1 Report
The paper discusses an interesting topic and proposes a new application of a KNN and Bidirectional Long Short Term Memory Model-based traffic prediction model.
In my opinion, however, the presentation of the paper could be improved, both in the methodology description and results presentation.
After a good introduction and presentation of adequate and updated literature, model description, and results presentation sections were not very smooth for me. My advice to the authors is to revise the paper by increasing the clarity of the proposed discussion.
The current text version also contains numerous editing and formatting errors, and the quality of the figures is poor and must be necessarily improved. In my opinion, the authors must also work on these aspects to bring the paper to the level required by the journal.
In general, my review advice is to provide a more straightforward explanation of the proposed methodology, deepening in section 2 the advantages of the new algorithm mentioned in the introductory parts.
I want to report some specific observations I would like to signal to the authors.
1. Regarding the two-way structure proposed and presented in section 2.2, in my opinion, it is necessary to explain the BILSTM structure and backward and forward transmission better, clarifying the difference between equations (8) and (9) and their average that is assumed as a result.
2. In general, it is intuitive that the results reported in terms of prediction are those on the 20% of the data that were not used in the training phase. Still, it would be helpful to specify this difference as often errors in the training phase and errors in the test phase are presented unclearly.
3. Figure 3 is a bit confusing. My advice is to number the sub-figures and insert an adequately explanatory caption. The quality of the graphs must be increased, which is currently not satisfactory.
4. Figure 4 also needs to be improved, as the labels are too small to read easily.
5. On page 9, there is a problem with the numbering of the figures: there are two figures 4. This problem and the one related to the progressive numbering of the following figures and references in the text must be resolved.
6. In the paper, there are two figures numbered 8, which, like the others, need improvements in graphic quality.
7. The conclusions appear to be too restricted. The authors should better highlight the work they have done and the results obtained and should recall the possible application uses of the algorithm to highlight the practical significance of the improvements achieved.
8. In my opinion, section 6, with the indications for future work, could be eliminated, reporting the content within the new conclusions.
I hope my comments can be useful to the authors to improve the presentation of their research work.
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.
all the details revised can be described in the file as follows.
Thanks you so much!
Author Response File: 
Author Response.docx
Reviewer 2 Report
Using data from the UK, the authors investigated different models of NN and develop one new approach for short-term traffic flow prediction. The implemented methodology is straightforward. The findings are reasonable. The paper is easily readable and understandable. The results seem to be useful for the practice.
Some minor comments:
Check throughout the paper with respect to citations (format) and references (tables and figures). Their some typos there.
Eqs.8 and 9, the right sides of both equations are identical. This must be an error.
Fig. 5, why using 200 sample data? It may be reasonable using 194 sample data for 2 whole days.
Figure 6, please give the corresponding detection points for X1-X14 (33..., 44... in fig. 2)
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.
all the details revised can be described in the file as follows.
Thanks you so much!
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
I thank the authors for considering my comments, which have been adequately addressed. The article is now suitable for publication, having adapted the formatting of the text, which has some minor editing and pagination problems.
