Research on the Influence Mechanism of Street Vitality in Mountainous Cities Based on a Bayesian Network: A Case Study of the Main Urban Area of Chongqing
Round 1
Reviewer 1 Report
The authors have been putting forward a study case to investigate the most important factors influencing the vitality of street vitality in mountainous cities. The different aspects the authors wanted to consider are being analysed using a methodology based on Bayesian networks to define the causal role of external interventions and to describe the causal relationships among multiple variables. To reduce the BN's accuracy, multicollinearity tests have been performed. The sensitivity analysis function has also been provided together with multifactor analysis.
The manuscript is well written, and the scientific soundness is respected, even though some minor spelling mistakes and blunders reduce the text quality.
The minor flaws that I am listing here should consider before the manuscript is being published:
line 14-17 - "In order to explore the influencing factors and the influencing mechanism of street vitality in mountain city. A new quantitative research method based..." the broken sentence is nonsense. Moreover, the word influencing is repeated.
- from lines 27 to 38. The important statement should be appropriately referenced.
- Line 45 - commas should be replaced with a full stop.
Line 47 - double commas
from Line 52 to 53 - That is an important statement, appropriate citations should be added.
Line 81 - check out the punctuation.
Line - 221 - 223 check out the punctuation.
Figure 4 - the colour palette should be improved if possible
Table 4 - if the first test has shown a severe multicollinearity condition for SVR and SCR, why is SCR mentioned in the second test results? At least, it is not that clear in the storytelling. Authors should consider that not all readers know BN's methodology.
Line 276 - the excessive emphasis should be avoided. If I were the author, I would not use outstanding referred to a performance of mine.
Figure 8 - the image resolution should be improved. The graphs are a little blurred
Line 337 - MHV should replace MVH
The References section looks not formatted according to the manuscript's body
that's all
Author Response
Response to Reviewer1 Comments:
The authors have been putting forward a study case to investigate the most important factors influencing the vitality of street vitality in mountainous cities. The different aspects the authors wanted to consider are being analysed using a methodology based on Bayesian networks to define the causal role of external interventions and to describe the causal relationships among multiple variables. To reduce the BN's accuracy, multicollinearity tests have been performed. The sensitivity analysis function has also been provided together with multifactor analysis.
The manuscript is well written, and the scientific soundness is respected, even though some minor spelling mistakes and blunders reduce the text quality.
The minor flaws that I am listing here should consider before the manuscript is being published:
Point 1. line 14-17 - "In order to explore the influencing factors and the influencing mechanism of street vitality in mountain city. A new quantitative research method based..." the broken sentence is nonsense. Moreover, the word influencing is repeated.
Response 1: We have revised the abstract and corrected the inappropriate wording.
Point 2. from lines 27 to 38. The important statement should be appropriately referenced.
Response 2: We have added references to the introduction from line 27 to 38.
Point 3. Line 45 - commas should be replaced with a full stop. Line 47 - double commas
Response 3: We have corrected the punctuation.
Point 4.from Line 52 to 53 - That is an important statement, appropriate citations should be added.
Response 4: We have added references to the introduction from line 52 to 53.
Point 5. Line 81 - check out the punctuation. Line - 221 - 223 check out the punctuation.
Response 5: We have corrected the punctuation.
Point 6. Figure 4 - the colour palette should be improved if possible
Response 6: We have adjusted the color of Figure 4 (now it is Figure 3).
Point 7.Table 4 - if the first test has shown a severe multicollinearity condition for SVR and SCR, why is SCR mentioned in the second test results? At least, it is not that clear in the storytelling. Authors should consider that not all readers know BN's methodology.
Response 7: In Table 4, the first test results showed that VIF of SVR and SCR were both greater than 10, indicating the existence of multicollinearity between the two. At this point, the SVR with the largest VIF1 value was removed for the second test. The second result showed that after SVR was removed, VIF2 of all other factors were less than 10, indicating that there was no serious multicollinearity among all other factors. The SVR of the second result in Table 4 was not analyzed as the result, and we modified the expression in the paper.
Point 8.Line 276 - the excessive emphasis should be avoided. If I were the author, I would not use outstanding referred to a performance of mine. 266
Response 8: We corrected the inappropriate wording.
Point 9. Figure 8 - the image resolution should be improved. The graphs are a little blurred
Line 337 - MHV should replace MVH.
Response 9: We have revised the mistake.
Point 10. The References section looks not formatted according to the manuscript's body
that's all.
Response 10: We checked the format of the references and modified it.
Reviewer 2 Report
The manuscript addresses major issue of street vitality in a mountainous city using Bayesian network. The authors have used great input parameters to train the model and they have archived better result as well. But I had difficulties with some sentences and felt a rigorous editing is required especially in introduction and method section. Also previous studies are not cited while explaining methods. I have listed few comments below which need further attention:
- Abstract, the motivation of the study is unclear, also the structure of abstract needs modification.
- Overall manuscript needs a thorough English grammar check.
- Line 98-100, The sentence is not making any sense. Please briefly define how bayesian networks let to a breakthrough in AI. I am also curious if there are any alternative to Bayesian networks? This manuscript doesn't tell us about other alternate methods or related studies.
- Section 2.1, Related references are missing in the paragraph
- Please include lat long information when you show a map. It is missing throughout the manuscript.
- Figure 7,8 and 9 are not clear. Replace it with a high resolution images.
Author Response
Response to Reviewer2 Comments:
The manuscript addresses major issue of street vitality in a mountainous city using Bayesian network. The authors have used great input parameters to train the model and they have archived better result as well. But I had difficulties with some sentences and felt a rigorous editing is required especially in introduction and method section. Also previous studies are not cited while explaining methods. I have listed few comments below which need further attention:
Point 1. Abstract, the motivation of the study is unclear, also the structure of abstract needs modification.
Response 1: We have revised the abstract and corrected the inappropriate wording.
Point 2. Overall manuscript needs a thorough English grammar check.
Response 2: We have embellished the paper.
Point 3. Line 98-100, The sentence is not making any sense. Please briefly define how bayesian networks let to a breakthrough in AI. I am also curious if there are any alternative to Bayesian networks? This manuscript doesn't tell us about other alternate methods or related studies.
Response 3: We modified it from line 105 to 119.
Point 4. Section 2.1, Related references are missing in the paragraph.
Response 4: We have added references to the introduction in Section 2.1.
Point 5. Please include lat long information when you show a map. It is missing throughout the manuscript. Figure 7,8 and 9 are not clear. Replace it with a high resolution images.
Response 5: We have updated the image (now they are Figure 6,7 and 8).
Reviewer 3 Report
The paper presents a Bayesian Network-GIS framework for assessing the urban vitality in Chongqing (China). The research aims to identify the factors influencing street vitality in mountainous cities.
The model core conceptualization is not clear: quantification and discretization of the street vitality (MHV), obtaining the Bayesian Network model through structure and parameter learning.
Furthermore, the particularities of the model for mountainous cities are insufficient emphasized, even though attributes for street elevation and slopes are used (probably with specific ranges in the case of Chongqing).
General comments
The abstract should be more concise and explicitly emphasize the particularities of the study. The term "mountain cities "is overused in some sentences (see line 13). And it is not sufficiently highlighted what procedures or categories of factors particularize the method for mountainous cities.
The introduction presents the main types of urban vitality approaches and the advantages of using big data and new technologies in assessing street vitality. Many statements seem taken from other papers not correctly quoted. E.g. Gehl 1987 (line 41), Gehl 2010 (lines 66, 173), Mehta 2007 (lines 66. 69. 183), etc. The references need to be attentively revised.
In my opinion, the Methodology section should be outlined accordingly to the steps of the research framework presented in the introduction (lines 84-94).
The VIF lower limit of 10 requires stronger argumentation for multicollinearity (lines 132-134). The reference Hair et all. 2006 is missing. Also, argumentation is required for the need of the factor discretization.
The street vitality indicator, denoted by MHV, was computed based on population activity density reflected by Baidu's LBS data (lines 169-171). Even though details about the used data are provided (lines 247-251), indications regarding the MHV computation are missing. Furthermore, the maps in Fig. 6 must be supplemented by discussing the range and distribution of MHV values.
Most of the influencing factors of street vitality (shown in Table 2) need clarifications on quantification methods (especially those based on areas, e.g., sky area). Some factors seem to be strongly dependent, e.g., street length and road density; POI density and bus stop density; but VIF values do not reflect the dependency in Table 4.
It is not clear how the factor discretization was applied.
The obtaining of the BN model through structure and parameter learning is not explained. But validation test values are presented for two categories of street vitality values (Table 5).
Additional comments
The definitions for Bayesian Networks need revisions." E" denotes a set of arcs (not only an arc), and" Teta" also denotes a set. It would be recommended a correlation between denotations and terms in Table 1 (section 2.1, lines 115-123).
The eqs. 1-3 (for F-score) are irrelevant and can be eliminated. In this research, F-score values are automatically computed by procedures included in the commercial software packages (section 2.1, lines 151-157).
The meaning of the ROC acronym should be provided. The sentence for the second parameter definition needs revision: the ROC curve cannot be a parameter (probably the area under the ROC curve) (line 158).
Fig. 2 shows a street cross-section, not a three-dimensional space (lines 225-226).
Author Response
Response to Reviewer 3 Comments
The paper presents a Bayesian Network-GIS framework for assessing the urban vitality in Chongqing (China). The research aims to identify the factors influencing street vitality in mountainous cities.
The model core conceptualization is not clear: quantification and discretization of the street vitality (MHV), obtaining the Bayesian Network model through structure and parameter learning.
Furthermore, the particularities of the model for mountainous cities are insufficient emphasized, even though attributes for street elevation and slopes are used (probably with specific ranges in the case of Chongqing).
General comments
Point 1.The abstract should be more concise and explicitly emphasize the particularities of the study. The term "mountain cities "is overused in some sentences (see line 13). And it is not sufficiently highlighted what procedures or categories of factors particularize the method for mountainous cities.
Response 1: We have revised the abstract and corrected the inappropriate wording. And we emphasized the difference between mountain city and plain city from line 78 to 83.
Point 2. The introduction presents the main types of urban vitality approaches and the advantages of using big data and new technologies in assessing street vitality. Many statements seem taken from other papers not correctly quoted. E.g. Gehl 1987 (line 41), Gehl 2010 (lines 66, 173), Mehta 2007 (lines 66. 69. 183), etc. The references need to be attentively revised.
Response 2: We have added references.
Point 3. In my opinion, the Methodology section should be outlined accordingly to the steps of the research framework presented in the introduction (lines 84-94).
Response 3: We have revised the methodology.
Point 4. The VIF lower limit of 10 requires stronger argumentation for multicollinearity (lines 132-134). The reference Hair et all. 2006 is missing. Also, argumentation is required for the need of the factor discretization.
Response 4: We have added references. And We have revised it in section 2.2.
Point 5. The street vitality indicator, denoted by MHV, was computed based on population activity density reflected by Baidu's LBS data (lines 169-171). Even though details about the used data are provided (lines 247-251), indications regarding the MHV computation are missing. Furthermore, the maps in Fig. 6 must be supplemented by discussing the range and distribution of MHV values.
Response 5: We have added the formula in the Appendix B. And We discussed figure 6 from line 275 to 281.
Point 6. Most of the influencing factors of street vitality (shown in Table 2) need clarifications on quantification methods (especially those based on areas, e.g., sky area). Some factors seem to be strongly dependent, e.g., street length and road density; POI density and bus stop density; but VIF values do not reflect the dependency in Table 4.
Response 6: Street length and street density, POI density and bus station density are indeed correlated to some extent, otherwise directed arcs cannot be formed between them through Bayesian networks. However, in this study, street length refers to the length between every two intersections, which can be understood as a street segment. This factor focuses on the street itself. Street density refers to the ratio of the sum of the road lengths in a certain area around a certain street segment to the area of this area. This factor is concerned with the environment around the street. Therefore, the above two factors are not as strongly dependent as expected, which can also be confirmed by the multicollinearity test results. POI contains a wide range of contents, including 16 categories of points, which more represent facilities such as catering, entertainment and life services, as shown in Figure 3. Therefore, there is no serious multicollinearity between POI density and bus stop density.
Point 7. It is not clear how the factor discretization was applied.
Response 7: We have revised it in section 2.2. And The discretization method is shown in Appendix A.
Point 8. The obtaining of the BN model through structure and parameter learning is not explained. But validation test values are presented for two categories of street vitality values (Table 5).
Response 8: We have revised it from line 304 to 311.
Additional comments
Point 9. The definitions for Bayesian Networks need revisions." E" denotes a set of arcs (not only an arc), and" Teta" also denotes a set. It would be recommended a correlation between denotations and terms in Table 1 (section 2.1, lines 115-123).
Response 9: " E" denotes evidence, which can refer to one node or multiple nodes.
Point 10. The eqs. 1-3 (for F-score) are irrelevant and can be eliminated. In this research, F-score values are automatically computed by procedures included in the commercial software packages (section 2.1, lines 151-157).
Response 10: We removed the eqs. 1-3 (for F-score) in line 174.
Point 11. The meaning of the ROC acronym should be provided. The sentence for the second parameter definition needs revision: the ROC curve cannot be a parameter (probably the area under the ROC curve) (line 158).
Response 11: We have revised it from line 182 to 183.
Point 12. Fig. 2 shows a street cross-section, not a three-dimensional space (lines 225-226).
Response 12: We removed Figure 2.
Reviewer 4 Report
Dear Authors,
The article concerns an important issue of the vitality of a city located in a mountain area.
Taking into account the large number of cities of this type and above all the issue of urban vitality itself as an important problem of urban planning, the article is an important contribution to further research on both issues.
The greatest advantage of the article is the methodology used and the results achieved thanks to it.
However, I have a few remarks related to the organization of the text.
First, I recommend that you separate the Introduction from the Research Review.
Introduction should be expanded. The same should be done with the Research Review. The Authors indicate several discourses, but not all of them are described sufficiently broadly. The topic is very broad. There are hundreds of publications on the subject, while the References chapter is not very extensive. I recommend treating this issue more thoroughly.
The purpose of the work should not be so far from the first sentence of the article.
The last sentence should be removed from the current version of the introduction. Describing the article structure is not necessary.
There is an evident disruption between the length of the Methodology and Results chapter. The Authors put a lot of work into the research, which I personally appreciate. I believe that the Land Readers can be informed of the results in a slightly longer narrative.
I also have comments on the Discussion and Conclusion chapter.
They should be separate. In the Discussion, it would be advisable to address two issues:
a) self-assessment, possible criticism and further possibilities of the methodological path used by the authors regarding the study of the vitality of (in) mountain cities,
b) references to world research in this area. To what extent do the Authors' research coincide with those previously carried out by other researchers, and to what extent are original. Readers of international scientific journals appreciate such a discussion. He writes it from an autopsy.
Figure 2 is not needed. It is illustrative only.
Sincerely
Author Response
Response to Reviewer 4 Comments
The article concerns an important issue of the vitality of a city located in a mountain area.
Taking into account the large number of cities of this type and above all the issue of urban vitality itself as an important problem of urban planning, the article is an important contribution to further research on both issues.
The greatest advantage of the article is the methodology used and the results achieved thanks to it.
However, I have a few remarks related to the organization of the text.
Point 1.First, I recommend that you separate the Introduction from the Research Review.
Introduction should be expanded. The same should be done with the Research Review. The Authors indicate several discourses, but not all of them are described sufficiently broadly. The topic is very broad. There are hundreds of publications on the subject, while the References chapter is not very extensive. I recommend treating this issue more thoroughly.
Response 1: We revised the introduction and added references to the introduction. In addition, we have added references to the methodology. Due to the length of the article, I am sorry that we did not add literature review separately.
Point 2.The purpose of the work should not be so far from the first sentence of the article.
The last sentence should be removed from the current version of the introduction. Describing the article structure is not necessary.
Response 2: We revised the introduction.
Point 3.There is an evident disruption between the length of the Methodology and Results chapter. The Authors put a lot of work into the research, which I personally appreciate. I believe that the Land Readers can be informed of the results in a slightly longer narrative.
I also have comments on the Discussion and Conclusion chapter.
They should be separate. In the Discussion, it would be advisable to address two issues:
a) self-assessment, possible criticism and further possibilities of the methodological path used by the authors regarding the study of the vitality of (in) mountain cities,
b) references to world research in this area. To what extent do the Authors' research coincide with those previously carried out by other researchers, and to what extent are original. Readers of international scientific journals appreciate such a discussion. He writes it from an autopsy.
Response 3: We separated the discussion from the conclusion and discussed the article according to your suggestion.
Point 4. Figure 2 is not needed. It is illustrative only.
Response 4: We removed Figure 2.
Round 2
Reviewer 3 Report
The manuscript was substantially improved.
Minor comments
1. Appendix B presents the calculation method of influencing factors that should precede Appendix A. Or Table 2, Appendix A and Appendix B could be concentrated into one coherent list.
2. In Appendix A, there is no point in listing values with so many decimal places. Unit measures should be specified for node variables/influencing factors.
3. In Appendix A, a value of MHV is missing. The MHV values need to be consistent with the lower and upper limits indicated in Figure 5 (right frame).
4. E denotes a directed arc between different variables (line 128). And then, E denotes evidence variables (line 333).
Author Response
Point 1. Appendix B presents the calculation method of influencing factors that should precede Appendix A. Or Table 2, Appendix A and Appendix B could be concentrated into one coherent list.
Response 1: When the paper introduces Table 2, it does not yet introduce the data. Instead of combined Table 2, Appendices A and B into one coherent list, we switched positions of Appendices A and B.
Point 2. In Appendix A, there is no point in listing values with so many decimal places. Unit measures should be specified for node variables/influencing factors.
Response 2: We have revised it and kept four decimal places for the whole paper.
Point 3. In Appendix A, a value of MHV is missing. The MHV values need to be consistent with the lower and upper limits indicated in Figure 5 (right frame).
Response 3: The discrete points of each influencing factor/node variable were originally listed in Appendix A, not the value range of each influencing factor/node variable. However, we updated Appendix A to present the value range of each influencing factor/node variable.
Point 4. E denotes a directed arc between different variables (line 128). And then, E denotes evidence variables (line 333).
Response 4: we have replaced E with the letter X as the referential letter of directed arc to avoid misunderstanding.
Reviewer 4 Report
Dear Authors,
I welcome the new, improved version of the article. I believe that it is now significantly more favorable for the reader.
I have two more minor comments.
The title of the article does not cause the conclusions to be split into two parts. I would avoid the subsections in the Conclusions.
Fig. 4 should also be enlarged to make its message clearer.
Author Response
Point 1. The title of the article does not cause the conclusions to be split into two parts. I would avoid the subsections in the Conclusions.
Response 1: We have removed the subsection title.
Point 2. Fig. 4 should also be enlarged to make its message clearer
Response 2: We replaced Fig. 4 with a clearer picture.
