An Unsupervised Machine Learning-Based Framework for Transferring Local Factories into Supply Chain Networks

Round 1

Reviewer 1 Report

The paper presents a framework based on hybrid PCA/K-means unsupervised machine learning for transferring local factories into supply chain networks applied in Malaysia. The paper is well presented. However, I encourage the authors to consider the following points for further consideration.

1-it is known that as the feature space is reduced through PCA, the loss metrics start to converge without significant impact on the accuracy measurement. I suggest to the authors to measure any metric such as the accuracy of their model in order to make a fair comparison with other model in the state-of-the-art in terms of system performance.

2-Make a comparison of the proposed technique (i.e., PCA and K-means method) with similar one in the literature to strength the analysis.

3-The paper contains some typos such as ‘Table 4.9’ in page 22 line 700.

Author Response

Comments and Suggestions for Authors

The paper presents a framework based on hybrid PCA/K-means unsupervised machine learning for transferring local factories into supply chain networks applied in Malaysia. The paper is well presented. However, I encourage the authors to consider the following points for further consideration.

1-It is known that as the feature space is reduced through PCA, the loss metrics start to converge without significant impact on the accuracy measurement. I suggest to the authors to measure any metric such as the accuracy of their model in order to make a fair comparison with other model in the state-of-the-art in terms of system performance.

• Yes, your comment is true. In the revised paper, the following steps are taken:
• 1- The significance of the features is compared using the Shapiro method. (kindly see page18)
• 2- A new section is added for calculating the score of the proposed method. For this purpose, we used silhouette and calinski-harabasz scores. (kindly see page 30)
• 3- A new section is added to compare the proposed method with a number of unsupervised methods that were frequently used in the literature. (kindly see page 23)

2-Make a comparison of the proposed technique (i.e., PCA and K-means method) with similar one in the literature to strength the analysis.

• Yes, many thanks for your constructive comment. In the revised paper, a new section (4.8) is added to compare the proposed method with a number of other methods that were mostly used before. (kindly see page 23)

3-The paper contains some typos such as ‘Table 4.9’ in page 22 line 700.

• The entire paper is revised again in terms of English level, the caption of tables and figures. (kindly see page 19)

Dear Reviewer,

On behalf of our team, I would like to thank you for the constructive comments you made, which helped us improve the level of our paper.

Warm regards,

Reviewer 2 Report

1. In the Abstract the authors should brief/highlight the actual methodology and practical application of the proposed work
2. Reference for specific values in Introduction part (line No. 28-33)
3. Reference Number [11] is presented without author name(s) even though other references are with authors and year Format.
4. In fig.4, “agglomerative” is having typo error.
5. Page 16,line No. 520 is having grammatical error
6. In the tables presented in 22, 23 the Name “Kuala Lumpur” has been typed with error
7. The hyperlink presented in page 20 has not been cited anywhere in that page (please refer MDPI Journal formatting procedure)
8. Similarly the “Map of Malaysia” is not given with proper reference in Fig. 15, page21, also Fig26, 27
9. The methodology section must be separately given to explain to understand “know-how”, “know-why” Questions..
10. The wholesalers’ best location at present scenario of Malaysia would have followed some strategies since they must be towards the development of their concerns. So the author must quote these strategies and compare with that of the present machine learning algorithm’s results.
11. It is also important to explain in what way these theoretical algorithms will be accepted and implemented by the real time manufacturers so as to improve the current state.
12. It is better to illustrate with a case study of one or two companies that are located in Malaysia.
13. In conclusion, the authors are expected to mention the practical applicability (in the selected companies) of the proposed work

Comments for author File: Comments.docx

Author Response

Comments and Suggestions for Authors

1. In the Abstract the authors should brief/highlight the actual methodology and practical application of the proposed work

• Yes, many thanks for your comment. In the revision, the abstract is revised and the important findings and highlights of the research are outlined. (Kindly see page 1)

1. Reference for specific values in Introduction part (line No. 28-33)

• Yes, the reference for the mentioned section is added. (Kindly see page 1)

1. Reference Number [11] is presented without author name(s) even though other references are with authors and year Format.

• Yes, in the revised paper, the reference is modified and updated. (Kindly see page 5)

1. In fig.4, “agglomerative” is having typo error.

• Figure 4 is removed due to the third reviewer’s comment.
• The entire paper is revised again in terms of typos and grammatical errors. All modifications are shown by Track Changes mode.

1. Page 16,line No. 520 is having grammatical error

• The entire paper is revised again in terms of typos and grammatical errors. (Kindly see page 13)

1. In the tables presented in 22, 23 the Name “Kuala Lumpur” has been typed with error

• Yes, the word is corrected. (Kindly see pages 17 and 19)

1. The hyperlink presented in page 20 has not been cited anywhere in that page (please refer MDPI Journal formatting procedure)

• Yes, your comment is true. In the revised paper, the cited hyperlink is modified based on the journal’s guideline (kindly see references 68 and 69, page 35).

1. Similarly the “Map of Malaysia” is not given with proper reference in Fig. 15, page21, also Fig26, 27

• The reference of the map of Malaysia is added to Figures 15, 26 and 27. (Kindly see pages 18 and 29)

1. The methodology section must be separately given to explain to understand “know-how”, “know-why” Questions.

• Yes, your comment is true. In the revised paper, a new section (3.1) is added describing the research methodology using a flow chart. (Kindly see pages 10 and 11)

1. The wholesalers’ best location at present scenario of Malaysia would have followed some strategies since they must be towards the development of their concerns. So the author must quote these strategies and compare with that of the present machine learning algorithm’s results.

• In the revised section, section 4.6 is revised explaining the settings that were considered in the machine-learning algorithm with the strategies of the owners of the systems. (Kindly see page 18)
• Moreover, a new paragraph is also added to compare the gained results with the predefined strategies. (Kindly see page 29)

1. It is also important to explain in what way these theoretical algorithms will be accepted and implemented by the real-time manufacturers so as to improve the current state.

• Yes, in the revision, a new paragraph was added indicating the reasons for using machine learning and explaining how to connect the outcomes of the research with the real-time data of the manufacturing firms. (Kindly see pages 29 and 30)

1. It is better to illustrate with a case study of one or two companies that are located in Malaysia.

• Yes, many thanks for your comment. Honestly, the outcomes are for a local chocolate company in Malaysia. But, due to security reason, we are not allowed to name their company or reveal more data. In the revised paper, we revised the section and added more details about the company. (Kindly see page 10)

1. In conclusion, the authors are expected to mention the practical applicability (in the selected companies) of the proposed work.

• Yes, in the revised paper, the conclusion section is updated by adding practical findings of the research along with more detailed results. (Kindly see pages 30 and 31)

Dear Reviewer,

On behalf of our team, I would like to thank you for the constructive comments you made, which helped us improve the level of our paper.

Warm regards,

Reviewer 3 Report

The article is interesting, but the authors absolutely need to shorten it by removing all the well known concepts of section 2 and 3 like. Why explaining supervised and unsupervised learning? Why providing diagrams about machine learning methods? It is not a book for an academic course in machine learning so, please, focus on the original contribution of this work (that, anyway, there are).

Author Response

The article is interesting, but the authors absolutely need to shorten it by removing all the well-known concepts of section 2 and 3 like. Why explaining supervised and unsupervised learning? Why providing diagrams about machine learning methods? It is not a book for an academic course in machine learning so, please, focus on the original contribution of this work (that, anyway, there are).

• Yes, many thanks for your comment. The entire paper is revised and the unnecessary parts of sections 2 and 3 (such as obvious information about machine learning algorithms etc.) were removed. (Kindly see pages 4 to 14)

Dear Reviewer,

On behalf of our team, I would like to thank you for the constructive comments you made, which helped us improve the level of our paper.

Warm regards,

Round 2

Reviewer 2 Report

Reviewer 3 Report

The authors have addressed my concerns.