A Geospatial Framework for Retail Suitability Modelling and Opportunity Identification in Germany

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Uploaded as a file.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

This is a bold attempt to combine a number of different approaches to find areas of greatest suitability for new retail stores in Germany. It combines a suitability analysis with spatial clustering and GWR to look at stationarity in the key variables across space. The author has a good understand of these techniques and applies them well but I have a number of major reservations.

First is the newness or uniqueness of the suitability analysis. The problem partly lies in the literature review which seems to not include the main papers on GIS use for retail analysis. The sources cited seem rather random to me and mainly belong to GIS in transport studies and land-use pattern analysis. The main GIS literature for retail analysis would identify a lot of work on suitability analysis already. For example, There exsit several references which combines data to provide ideal zones (the author uses the term priority grid cells but its largely the same) and suitability analysis undertaken by data overlay in  a GIS. You can consider supplementing your introduction with key references from this research area (the example of suitability analysis for pawn shops in Houston is very similar). I don’t also agree with the assertion that there are no rural accessibility studies or country-wide analysis. It is there if you look for it. (there is indeed a long and varied literature on retail location analysis which is largely missing here).

Second, the results seem very obvious for the suitability and clustering parts. The clusters are all in areas of high population density which is not surprising! This is despite the early comments criticising studies for focusing on urban areas only. The problem is accentuated by the poor maps – I think these are maps! Its impossible to know where these high suitability clusters are in Germany and if any of these are in rural areas? Also the small representation of the clusters means it is impossible to see areas of high yellow and red shading. So we need better underpinning maps, which includes the actual geography of Germany, and a much more detailed discussion of the geography of the results.

Third, the results from the GWR are interesting but what is the behavioural explanation for these results? – why is population density (seen to be universally important) or age of residents more important in some areas compared to others? For the GWR results to be useful we need to understand better why these coefficients vary over space.

So overall I need to be convinced this is a novel approach to retail site location. There may be a useful paper around more detailed discussion of the results (such as where the analysis provides some surprises spatially) but its not there yet.

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript titled “A Geospatial Framework for Retail Suitability Modeling and Opportunity Identification in Germany” developed an integrated geospatial framework specifically incorporating spatial autocorrelation and geographically weighted regression in the context of examining retail mapping and modeling in Germany.

Overall, the manuscript is well-organized and that the language is clear and concise. The figures and tables add useful support to the arguments presented in the work. However, there are some specific elements of the work that needs to be addressed as follows:

01. The background literature review is not comprehensive. As a result, dated references are used to justify certain claims (such as on page 2, line 46 “Recent advances in geographic information systems…..”). Also, there is a challenge to identify the real impact of the research work presented and to understand how it builds on what already exists in the published literature.

02. There are no explicit goal or objectives for the research work other than on page 2line 76 where it states that “This study adds to the literature on retail site selection by combining detailed demographic data…”. A more comprehensive literature review in Section 2 is needed to justify the statement. In addition, the developed framework in Figure 1 is a standard workflow process and more needs to be added to show how it is actually unique or novel.

03. In Figure 1, there is a “1” label in the last arrow and it is not clear what that is signifying.

04. It is stated that the 2011 German census data was used (page 6 line 225). Why wasn’t a more recent census used to gain understanding of more recent processes and dynamics. Also, would the 1km2 grid cell size be sufficiently granular to understand the local patterns?

05. On page 6, line 229 it would be useful to add that the EPSG:3035 is a coordinate system based on the Lambert Azimuthal Equal Area projection using the ETRS89 geodetic data.

06. On page 7, line 257 it was mentioned that expert knowledge was used to find a suitability score (from assigning suitable weights to w1, w2, w3, w4 and w5). Who were the experts used in the analysis and where they a representative sample? In line 286 it was mentioned that “I assigned weights….” seeming to suggest that the author assigned that the weights as a single expert.

07. On page 8 line at 308, the value of k=5 was used. What is the justification for this value?

08. On page 8 line 334. It was mentioned that the sensitivity analysis was done by generating alternative scenarios of various weight combinations. It would be useful to somehow summarize these scenario values and present them as a table or figure.

09. On page 8 line 339 the “$” symbol in the Pearson value should be removed.

10. On page 9 line 369 it was mentioned that shop density was obtained/calculated. How was this density calculated or was it downloaded directly from OpenStreetMaps?

11. On page 9 lines of 373 to 375 a suitability score value of greater than 1.5 was chosen. What is the justification for this value?

12. In Table 2 and Table 3, the values need to be reported to a consistent number of decimal places. For example, in Table 2 population and %Women has numbers with varying precision.

13. In Figure 2, some suitability scores are negative. Earlier in the manuscript it was mentioned that suitability scores range from 1 to 5. Why are there negative suitability scores?

14. In Figure 7, the shape of the study area seems to be distorted in (a) and (b) compared to (c).

15. On page 18 line 556. It is claimed that “This study offers a novel, spatially explicit framework…” which cannot be strongly supported given that the detailed background literature review was not done.

Overall, the manuscript can make a useful contribution. However, it needs to situate itself more clearly in the existing published knowledge base about spatially explicit (GIS) tools and how these have been used and/or adapted for retail analysis. Furthermore, careful attention needs to be given on the reporting of some of the research choices with solid justifications.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

Comments and Suggestions for Authors

This study introduces an integrated geospatial framework for retail opportunity mapping in Germany by combining multi-criteria suitability modeling with spatial autocorrelation analysis and geographically weighted regression (GWR). The study is innovative and the conclusions have practical applications. However, there is still a lot of room for improvement in the article, and the specific issues are as follows:

1. Research basis and positioning of selected topics.

The research is not innovative enough and lacks breakthrough contributions. The remote sensing and deep learning methods used in the article have been explored in more literature, and lack substantial improvement or adaptive optimization of existing methods, so the overall innovation is not outstanding. In addition, the research motivation and application value are not fully developed. For example, the introduction part does not clearly state the practical significance of the research in actual geospatial decision-making or urban management, and lacks the docking with policy, social or technical needs, which reduces the real value of the research.

2. Theoretical framework and methodological design.

The logic of the method is not clear enough, and the model selection lacks theoretical support. The text lacks a systematic description of the structure selection and parameter setting of the deep learning model used, and fails to account for why a specific network architecture was chosen, making the technology selection lack a theoretical basis. Comparison methods are not sufficiently set up. Although the article carries out some algorithm comparisons, it lacks systematic comparisons of baseline methods or representative remote sensing segmentation methods, which cannot adequately explain the advantages of the methods used.

3.Data and experimental part.

The representativeness of the sample set is doubtful. The article does not provide a clear description of the background information of remote sensing image data, such as the collection area, time, and terrain category, which limits the reproducibility and generalization ability assessment of the experimental results.

The experimental evaluation index design is single. The evaluation mainly focuses on traditional indicators such as IoU, precision, recall, etc., and lacks richer remote sensing segmentation quality assessment dimensions such as spatial continuity, edge precision and robustness.

4.Result analysis and visualization expression

The result interpretation lacks depth. Although the image segmentation results are displayed, there is a lack of spatial interpretation or geographic meaning analysis of the patterns behind the results, which fails to reflect the depth of interpretation that geoinformatics should have. There is redundancy and lack of information in the graphical design. Some of the graphs are repetitive, while the comparison images or error analysis graphs are missing, and the graph design needs to be optimized.

5、Conclusion and future research

The conclusion statement is too simple and generalized. The conclusion fails to clearly state the main contributions, technical limitations and possible expansion directions of the study in the light of the experimental findings, which affects the overall completeness of the article. There is a lack of constructive guidance for future research. Suggestions such as “the algorithm can be optimized in the future” and “it can be used in more areas” are superficial, and there is a lack of specific ideas on method expansion, data integration, or application in real-world scenarios.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

First I congratulate the author for undertaking a major rewrite so quickly and thoroughly and a much more impressive manuscript has emerged. The maps are a major improvement. I also think the literature review is much more appropriate. My final and only remaining reservation is around Fig 9 and its discussion - all the action takes place around Hamburg (so i am still not wholly convinced the GWR results have been explained - why so much variation within Hamburg compared to say Berlin?) and perhaps it would be useful to map that city only (i.e zoom into Hamburg only) so that the reader can see the spatial variations more clearly and the discussion could refer to that more local map- its the last of the original maps to survive and I think it need to be revamped like the others.  Other than that I am happy to recommend publication - again , well done author!!

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you very much for addressing all the feedback and the detailed explanations you have provided to clarify the points. The manuscript is now significantly improved and will make a valuable contribution to the knowledge base in this research area.

I have a few minor suggestions as you finalize the manuscript:

• The Abstract text should read as one flowing paragraph rather than broken into three separate paragraphs as currently exists
• In various parts of the manuscript, especially Section 3, you wrote in the first person "I". For example you stated "I accessed these datasets using the...". With this language style is comes across as a technical report rather than as a scientific paper. Ideally, you might want to consider changing this style to an inactive voice, for example: "These datasets were accessed using the...". Another example is: "For the spatial suitability analysis, I used a weighted multi-criteria approach built with the terra package in R" can become "A weighted multi-criteria approach built with the terra package in R was used for the spatial analysis."

Author Response

Thank you very much for your thoughtful feedback and for recognizing the improvements made to the manuscript. I have carefully implemented your suggestions:

• The Abstract has been revised to flow as one continuous paragraph.

• All instances of first-person language have been rephrased into an inactive voice to ensure a more scientific tone throughout the manuscript.

I appreciate your guidance, which has been very helpful in refining the paper.

Reviewer 4 Report

Comments and Suggestions for Authors

The comments raised by the reviewers have been well revised, and the quality of the article has been greatly improved. The article can now be accepted.

Author Response

Thank you very much for your positive feedback and for recognizing the revision. I sincerely appreciate your valuable comments and suggestions, which have helped to significantly improve the quality of the article. I am grateful for your time and consideration in reviewing the manuscript.