Dynamic Lead-Time Forecasting Using Machine Learning in a Make-to-Order Supply Chain

Round 1

Reviewer 1 Report

This study uses traditional machine learning methods to investigate delivery times' dynamic prediction and validate them with real-world data. It is a fascinating study, and the results are valid for practical applications. However, the problem-solving techniques used are not innovative, and the article is not very readable and needs substantial revision.

Some issues for the authors' reference.

1. On pages 20-21, it is not appropriate to use type I error for accuracy assessment, and it is suggested to use common indexes such as RMSA and MSE.
2.  A systematic description of the proposed method is suggested, such as a graphical flow approach.
3. The adopted machine learning method is not described in detail.
4. Why traditional linear regression or logistic regression is used? It is recommended to compare at least five similar methods. 
5. The authors do not clearly understand the basic ideas of machine learning, such as training samples, testing samples, and especially do not see the specific effects of "learning."
6. Figure 2 should provide statistical analysis results. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper concentrates on the important issue from the perspective of enterprises (especially the logistics companies) in regard to the dynamic forecasting for the purpose of shipment temporal consolidation optimization.

However, before publication the paper needs following improvements:

1. Motivation is weak. Authors could better define the aim of the paper. Moreover, they could better point out the existing gap in the literature.
2. In “Introduction” section, the main research ideas concluded by international researches should be express. The main aim of the paper should be included in the methodology part. In introduction, the general aim of the research should be explained.
3. Introduction should be extended, i.e. brief information on used methods, why this method was chosen instead of etc.
4. Literature review and improvements in regard to literature background should be made. Examples of articles that may be useful for the authors identifying similar contributions and better show what is the specific contribution.

• Aylak, B.L.; İnce, M.; Oral, O.; Süer, G.; Almasarwah, N.; Singh, M.; Salah, B. Application of Machine Learning Methods for Pallet Loading Problem. Sci. 2021, 11, 8304. https://doi.org/10.3390/app11188304
• Kosasih, Edward Elson; Brintrup, Alexandra. (2021) A machine learning approach for predicting hidden links in supply chain with graph neural networks. International Journal of Production Research, p. 1-14. DOI: 10.1080/00207543.2021.1956697
• Lewandowska, A. (2021) Interactions between investments in innovation and SME competitiveness in peripheral regions. Polish case study. Journal of International Studies, 14(1), 285-307. doi: 10.14254/2071-8330.2021/14-1/20
• Hanbazazah, Abdulkader S.; Castro, Luis E.; Erkoc, Murat; Shaikh, Nazrul I. (2020) In-transit freight consolidation of indivisible shipments. Journal of the Operational Research Society, 71(1), 37-54. 18p. DOI: 10.1080/01605682.2018.1527191.

In my opinion these four references cannot be omitted.

5. Discussion should be better. The results are presented clearly, but the findings are not compared and contrasted with relevant literature. Linking theoretical considerations with empirical findings and providing some planning insights is critical in a journal with the scope of Applied Sciences.
6. The limitations of the research should be given in the discussion or conclusions.

I hope that my comments are helpful to you as you continue your work on this project. Good luck.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The author actively revises the issues raised and gives sufficient explanation to meet the criteria for acceptance of the paper.