Next Article in Journal
Research on Identification Method for Interface Flange in Automatic Docking System of Fluid Loading and Unloading Arm for Bottom Loading
Next Article in Special Issue
Special Issue: Smart Service Technology for Industrial Applications
Previous Article in Journal
Dual Band Electrically Small Complementary Double Negative Structure Loaded Metamaterial Inspired Circular Microstrip Patch Antenna for WLAN Applications
Previous Article in Special Issue
Multi-Relational Graph Convolution Network for Service Recommendation in Mashup Development
 
 
Article
Peer-Review Record

Applying ANN and TM to Build a Prediction Model for the Site Selection of a Convenience Store

Appl. Sci. 2022, 12(6), 3036; https://doi.org/10.3390/app12063036
by Hsin-Pin Fu 1, Hsiao-Ping Yeh 1, Tein-Hsiang Chang 2,*, Ying-Hua Teng 1 and Cheng-Chang Tsai 1
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Appl. Sci. 2022, 12(6), 3036; https://doi.org/10.3390/app12063036
Submission received: 31 December 2021 / Revised: 17 February 2022 / Accepted: 13 March 2022 / Published: 16 March 2022
(This article belongs to the Special Issue Smart Service Technology for Industrial Applications)

Round 1

Reviewer 1 Report

Thanks for letting me review this interesting paper.

This manuscript is a well-organized paper, and it is easy to follow the history behind it. In my opinion, the ideas constitute marginal contributions, backed up by empirical findings. As a whole, we are in front of a promising paper. Nevertheless, there are some claims I disagree with; a discussion about them would benefit the manuscript.

Here I develop a question/answer self-interview to address the paper relevance and justify this recommendation:

1. Does the article fit the scope of the journal?
A: Yes, the article is closely related to the scope of this journal.

2.    Is the research novel?
o    A: I have not seen the TM in this context. But Retail Store Location is not new. Other authors have already addressed the inclusion of consumers and profit metrics. Please recognize the body of literature that focuses on that field (some examples at the end of this review).  
 
3.    Is the title representative of the article contents?
o    A: Yes, it is.

4.    Does the abstract summarise the contents?
o    A: Yes, it does.

5.    Is the state-of-the-art well described and the knowledge gap clearly defined?
o    A: I believe it is not. Please review the recommended literature (see 2.) and perhaps some related branches. Please point out the approach's limitations and indicate the gap in the literature that the authors are bridging with this manuscript.

6.    Are the objectives well-articulated?
o    A: Yes, they are.

7.    Is the applied research methodology solid?
o    A: As in many other ANN approaches, It is hard to produce a research paper addressing the wide variety of models and strategies. Nevertheless, as presented, I believe the research methodology clearly states the challenge, and the authors appropriately narrowed their proposal and innovatively used TM. Thus, it is appropriate but inherits the characteristics of ANN-related research.

8.    Are the results reliable, and have the objectives been reached?
o    A: In my opinion, the authors considered empirical results, and I have no evidence of another approach (please provide comparisons). It would be nice to have access to more insights about the data, links, and perhaps some more information to allow colleagues following this research line to reproduce the results, and most importantly, compare against a benchmark.

9.    Are the limitations correctly mentioned?
o    A: I believe they are not. Please see the previous comments and some particular suggestions at the end of this review.

10.    Are the conclusions justified?
o    A: Yes, they are detailed. It would be nice to comment on further research and potential applications.

It would be nice to have:
It would be nice to have an analysis and prospective proposals about the applications of the model and its relationships with literature regarding other kinds of applications; I believe there are efforts in marketing research that are closing the gap between retail and location decisions.

In the same line, it would be nice to have rough estimations about potential advantages that companies could reach through storing this data and using your framework.

Some specific comments:
A) In the Introduction: the author claims that

"... By reviewing past literature of site selection, there were many literatures using operational research method based on cost minimization to decide the industrial site location but few studies of site location decision of revenue maximum on retailing. ..."

I'm afraid I have to disagree because there is a complete body of literature on retail store location with profit-driven metrics and, of course, focused on costs and consumers' experience, preferences, and quality of service. Please see, for example:

https://doi.org/10.1016/j.apm.2019.05.040
https://doi.org/10.1016/j.apgeog.2018.08.007
H. R., Ganesha, and Aithal, P. S. and P, Kirubadevi, Ideal Store Locations for Indian Retailers – An Empirical Study (May 12, 2020). International Journal of Management, Technology, and Social Sciences (IJMTS), 5(1), 215-226 (2020)., Available at SSRN: https://ssrn.com/abstract=3611474
https://link.springer.com/chapter/10.1007/978-3-319-95162-1_27
https://ieeexplore.ieee.org/document/9548070
https://doi.org/10.1080/23754931.2018.1527720

B) Section 3 would benefit from leaving the description of general and well-known methodologies to the specific proposal of the authors in which they applied those methods, and more importantly, hopefully, a general framework for readers, colleagues, and practitioners, which allows and helps to use this proposal better.

C) Section 3.1, lines 143 to 147. May you please explain how your methodology is dealing with overfitting? How generalizable is what you are proposing?

Recommendation
I recommend revising this paper; please drop lines to discuss some of the comments and then resubmit.

Author Response

Dear reviewer 1:

The revised manuscript had been revised according to your valuable comments.

Fu

----

Reviewer 1

This manuscript is a well-organized paper, and it is easy to follow the history behind it. In my opinion, the ideas constitute marginal contributions, backed up by empirical findings. As a whole, we are in front of a promising paper. Nevertheless, there are some claims I disagree with; a discussion about them would benefit the manuscript.

Here I develop a question/answer self-interview to address the paper relevance and justify this recommendation:

  1. Does the article fit the scope of the journal?
    A: Yes, the article is closely related to the scope of this journal.

 

  1. Is the research novel?
    A: I have not seen the TM in this context. But Retail Store Location is not new. Other authors have already addressed the inclusion of consumers and profit metrics. Please recognize the body of literature that focuses on that field (some examples at the end of this review).  
    To the Reviewer:

The papers you recommended are added in the section of Literature Review.

 

  1. Is the title representative of the article contents?
    A: Yes, it is.

 

  1. Does the abstract summarise the contents?
    A: Yes, it does.

 

  1. Is the state-of-the-art well described and the knowledge gap clearly defined?
    A: I believe it is not. Please review the recommended literature (see 2.) and perhaps some related branches. Please point out the approach's limitations and indicate the gap in the literature that the authors are bridging with this manuscript.

To the Reviewer:

More paragraphs are added in the section of Literature Review.

 

  1. Are the objectives well-articulated?
    A: Yes, they are.

 

  1. Is the applied research methodology solid?
    A: As in many other ANN approaches, It is hard to produce a research paper addressing the wide variety of models and strategies. Nevertheless, as presented, I believe the research methodology clearly states the challenge, and the authors appropriately narrowed their proposal and innovatively used TM. Thus, it is appropriate but inherits the characteristics of ANN-related research.

 

  1. Are the results reliable, and have the objectives been reached?

A: In my opinion, the authors considered empirical results, and I have no evidence of another approach (please provide comparisons). It would be nice to have access to more insights about the data, links, and perhaps some more information to allow colleagues following this research line to reproduce the results, and most importantly, compare against a benchmark.

To the Reviewer:

Data analytical procedures of BPN and TM are added in the sections of Case study and Methodology and the corresponding references are also cited in order for readers of this paper to reproduce the analyses applying on their data.

 

  1. Are the limitations correctly mentioned?

A: I believe they are not. Please see the previous comments and some particular suggestions at the end of this review.

To the Reviewer:

A paragraph in regard to research limitation is added in the section of Conclusions.

 

  1. Are the conclusions justified?

A: Yes, they are detailed. It would be nice to comment on further research and potential applications.

It would be nice to have:

It would be nice to have an analysis and prospective proposals about the applications of the model and its relationships with literature regarding other kinds of applications; I believe there are efforts in marketing research that are closing the gap between retail and location decisions.

In the same line, it would be nice to have rough estimations about potential advantages that companies could reach through storing this data and using your framework.

To the Reviewer:

A paragraph in regard to implications, further works, and research limitation is added in the section of Conclusions.

 

Some specific comments:

  1. A) In the Introduction: the author claims that

"... By reviewing past literature of site selection, there were many literatures using operational research method based on cost minimization to decide the industrial site location but few studies of site location decision of revenue maximum on retailing. ..."

To the Reviewer:

This controversy sentence has been deleted in order not to mislead readers.

 

I'm afraid I have to disagree because there is a complete body of literature on retail store location with profit-driven metrics and, of course, focused on costs and consumers' experience, preferences, and quality of service. Please see, for example:

https://doi.org/10.1016/j.apm.2019.05.040

https://doi.org/10.1016/j.apgeog.2018.08.007

  1. R., Ganesha, and Aithal, P. S. and P, Kirubadevi, Ideal Store Locations for Indian Retailers – An Empirical Study (May 12, 2020). International Journal of Management, Technology, and Social Sciences (IJMTS), 5(1), 215-226 (2020)., Available at SSRN: https://ssrn.com/abstract=3611474

https://link.springer.com/chapter/10.1007/978-3-319-95162-1_27

https://ieeexplore.ieee.org/document/9548070

https://doi.org/10.1080/23754931.2018.1527720

 

  1. B) Section 3 would benefit from leaving the description of general and well-known methodologies to the specific proposal of the authors in which they applied those methods, and more importantly, hopefully, a general framework for readers, colleagues, and practitioners, which allows and helps to use this proposal better.

To the Reviewer:

The operating procedure of TM and TM are added in the Section 3.2. The reference is also cited.

 

  1. C) Section 3.1, lines 143 to 147. May you please explain how your methodology is dealing with overfitting? How generalizable is what you are proposing?

To the Reviewer:

Thanks for your valuable comments. There is no overfitting problem. Please see 4.4 section in blue and Fig. 3.

 

Recommendation

I recommend revising this paper; please drop lines to discuss some of the comments and then resubmit.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper entitled "Applying ANN and TM to Build the Prediction Model for Site Selection of a Convenience Store" offers an artificial neural network application combining back-propagation neural network (for location prediction) with Taguchi experiment design (for defining the optimal parameters). It is the application of two traditional tools, not much innovative, but relevant in finding the most suitable site for Convenience Stores, and it can be reproduced in other sectors of economic activity. Please address my comments for extending the paper.

 

Introduction:

  1. “By reviewing past literature of site selection, there were many literatures
    using operational research method based on cost minimization to decide the industrial site location but few studies of site location decision of revenue maximum on retailing”
    – This is a strong assertion that requires a reference.
  2. Please correct the sentence have also been used effectively, for example. to determine wind farm…”

 

Literature Review:

  1. Literature should be improved. The paper reports only 18 references, of which 9 are commented in the literature review section. It transmits the idea that the work is not theoretical based enough, or the idea that the authors did not take the attention to investigate the topic well adequately to state that there are only “few studies of site location decision of revenue maximum on retailing”. You can find in MDPI journals plenty of literature and applications on Backpropagation Neural Networks and Location Analysis. I would suggest a table summarizing the main literature contributions and differences from the proposed approach, but feel free to address the way you feel appropriate. I recommend expanding (not only) this section by including 10 to 20 more references.

 

Methodology

  1. The audience would appreciate a more detailed methodology. Notations and indexes are missing; there is no explanation on whether bias has been considered or how the training instance on CVSs locations would nudge the weights and biases; no comments about the considered transfer function; there are just a few comments on the back-propagation neural network illustration (and it should be placed after the explanation)… The methodology's essence is how changes in the parameters would produce the most rapid increase to the revenue (or reduction in the cost), and this is not clearly expressed.
  2. Please include a paragraph stating how the Taguchi method was applied in your simulation, and please cite the seminal work of Genichi Taguchi on loss function and experiment design.

 

Case Study

  1. How was data collected? Field interview? Questionaries? Is a formal database available online? How was the turnover success parameter defined?
  2. Please provide additional comments on the higher differences between prediction and actual daily turnover. I think this discussion is crucial.

 

Conclusion

  1. Include one paragraph or two on the managerial implications that should be addressed from the optimal CVS location methodology, additional limitations on the potential factors that were not considered and other restrictions and proposals for extensions.

Author Response

Dear Reviewer 2

The revised manuscript had been revised according to your valuable comments.

Fu

---

Reviewer 2

The paper entitled "Applying ANN and TM to Build the Prediction Model for Site Selection of a Convenience Store" offers an artificial neural network application combining back-propagation neural network (for location prediction) with Taguchi experiment design (for defining the optimal parameters). It is the application of two traditional tools, not much innovative, but relevant in finding the most suitable site for Convenience Stores, and it can be reproduced in other sectors of economic activity. Please address my comments for extending the paper.

 

Introduction:

  1. “By reviewing past literature of site selection, there were many literatures
  2. using operational research method based on cost minimization to decide the industrial site location but few studies of site location decision of revenue maximum on retailing” – This is a strong assertion that requires a reference.
  3. Please correct the sentence have also been used effectively, for example. to determine wind farm…”

To the Reviewer:

The typo is corrected.

 

Literature Review:

  1. Literature should be improved. The paper reports only 18 references, of which 9 are commented in the literature review section. It transmits the idea that the work is not theoretical based enough, or the idea that the authors did not take the attention to investigate the topic well adequately to state that there are only “few studies of site location decision of revenue maximum on retailing”. You can find in MDPI journals plenty of literature and applications on Backpropagation Neural Networks and Location Analysis. I would suggest a table summarizing the main literature contributions and differences from the proposed approach, but feel free to address the way you feel appropriate. I recommend expanding (not only) this section by including 10 to 20 more references.

To the Reviewer:

More paragraphs with more references are added, in read, in the section of Literature Review.

 

Methodology:

  1. The audience would appreciate a more detailed methodology. Notations and indexes are missing; there is no explanation on whether bias has been considered or how the training instance on CVSs locations would nudge the weights and biases; no comments about the considered transfer function; there are just a few comments on the back-propagation neural network illustration (and it should be placed after the explanation). The methodology's essence is how changes in the parameters would produce the most rapid increase to the revenue (or reduction in the cost), and this is not clearly expressed.

To the Reviewer:

Thanks for your valuable suggestions. To explain how the training instance would affect the robustness of the proposed method, different numbers of training instances were used to assess the performance of the ANN-TM prediction models. The results were evaluated based on four considered numbers of training examples, 16, 32, 48, and 64, respectively. The description and explanation have been included in the section 4.5. Also, the parameter design is very serious (we added red in see 4.3 and Table 10).

 

  1. Please include a paragraph stating how the Taguchi method was applied in your simulation, and please cite the seminal work of Genichi Taguchi on loss function and experiment design.

To the Reviewer:

The operating procedure of TM is added in the Section 3.2. The reference is also cited.

 

  1. How was data collected? Field interview? Questionaries? Is a formal database available online? How was the turnover success parameter defined?

To the Reviewer:

The data used in this study are secondary data, with approval, provided by one of CVS company in Taiwan. In this paper, one manager, in the case company, who was accountable to the retail site decision-making is in acknowledgement of data provision and consultancy.

 

  1. Please provide additional comments on the higher differences between prediction and actual daily turnover. I think this discussion is crucial.

To the Reviewer:

The comments on the big differences between predicted and actual turnovers is added in the last paragraph of the section 4.1.

 

Case Study:

None.

 

Conclusion:

  1. Include one paragraph or two on the managerial implications that should be addressed from the optimal CVS location methodology, additional limitations on the potential factors that were not considered and other restrictions and proposals for extensions.

To the Reviewer:

Managerial implications, research limitations, and further works are all added in the section of Conclusions.

 

Author Response File: Author Response.pdf

Back to TopTop