Bridging Accuracy and Interpretability: A Decision Support System for Stock Deployment and Additive Manufacturing Decisions in Spare Parts Distribution Networks

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The topic fits well within the scope of Logistics and contributes to the growing literature on data-driven decision support systems (DSSs) for supply chain configuration, particularly in the context of additive manufacturing. The manuscript is generally well structured, methodologically sound, and clearly written. The comparison with the prior decision-tree-based DSS by Cantini et al. is appropriate and highlights the value of the proposed approach in terms of predictive accuracy and interpretability. However, conceptual, methodological, and presentation-related issues should be addressed before the manuscript can be considered for publication.

1. As I understand it, the entire DSS is trained on data generated exclusively from a single reference mathematical model (Cantini et al. [30]). I think the authors should be more explicit about the external validity of the DSS. In particular, when alternative cost structures, inventory policies, or demand processes are considered.

2. The paper emphasizes the improvement from 77% (decision tree) to 93.4% (random forest). However, the managerial significance of this improvement is not sufficiently discussed.

3. The terms “DN configuration”, “inventory deployment policy”, and “DN replacement” are sometimes used interchangeably. Consider harmonizing the terminology to avoid confusion.

Author Response

Please, consult the attached file. Thank you.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Dear author,
Kindly include the following in the revised submission.

• Abstract (Lines 12–15)
  The abstract outlines the study aim clearly; however, the methodological approach is not sufficiently indicated.

• Abstract (Lines 18–20)
  Quantitative results are reported without specifying the evaluation criteria or baseline for comparison.

• Keywords (Line 24)
  Some keywords are broad and may be refined to better reflect the analytical focus of the study.

• Introduction, Para 1 (Lines 32–38)
  The background is relevant, but recent studies from the last 3–4 years could be better integrated.

• Introduction, Para 3 (Lines 54–58)
  The research gap is implied rather than explicitly stated and should be articulated more clearly.

• Methodology Section (Lines 79–96)
  The overall workflow is understandable; however, justification for selected parameters is limited.

• Data Description (Table 1, Lines 101–108)
  The table is informative, though units and variable definitions should be consistently presented.

• Model Development (Lines 122–140)
  The modeling approach is appropriate, but assumptions made during formulation require clarification.

• Results and Discussion, Para 2 (Lines 168–176)
  Observed trends are discussed adequately; linking them more explicitly to prior literature would strengthen the analysis.

• Figure 4 (Lines 181–185)
  The figure supports the discussion, though axis labels and legends can be improved for readability.

• Sensitivity Analysis (Lines 198–206)
  The analysis is relevant but briefly presented; key influencing parameters should be highlighted more clearly.

• Limitations Section (Lines 214–218)
  Limitations are acknowledged; however, their implications on result interpretation could be discussed more concisely.

• Conclusion (Lines 230–238)
  The conclusions summarize findings well but may be shortened to emphasize the principal outcomes.

• References (Lines 245–260)
  Some references are dated; inclusion of more recent and high-impact sources is recommended.

Comments on the Quality of English Language

The English could be improved to more clearly express the research.

Author Response

Please, consult the attached file. Thank you.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Thanks for submitting this manuscript. This paper investigates the integration of Random Forest models with SHAP-based Explainable Artificial Intelligence to support decision-making in spare parts distribution networks. By focusing on the choice between centralized, decentralized, and hybrid stock deployment combined with additive or conventional manufacturing, the authors aim to bridge the gap between high accuracy and interpretability. The main contribution lies in providing a Decision Support System that achieves a 93.4% accuracy rate while offering visual explanations for the underlying logic of the recommendations.

I encourage you to evaluate the following observations with a critical perspective and reflect them in your revised manuscript.

- In the introduction, you state “Despite the potential of this combined approach, no existing studies have applied ML together with post-hoc XAI to spare parts stock deployment decision-making.”. To prove this claim, I strongly suggest you incorporate a comparison table. This table should allow the reader to see at a glance that other papers in the literature have specific gaps which your study fills for the first time in absolute terms. This visual aid will make your contribution emerge with much more intensity and authority.

- Figure 1 is currently too small and crowded; I suggest you significantly increase its size or simplify its layout to improve readability, as the current version makes it difficult to distinguish key details.

- Regarding the literature search described in Section 2.2, you explain the queries used but you fail to mention the exact date when these searches were performed. For the sake of methodological correctness and transparency, you should insert the date of the search to define the boundaries of your state-of-the-art analysis. 

- Furthermore, the current structure of the paper is somewhat disjointed. I find that Chapter 3, which provides the decision-making problem description, would be much more effective if moved as the first subparagraph of the Materials and Methodology section. This change will allow the reader to understand the problem before being introduced to the technical solution, creating a more logical and standard academic flow.

- Regarding Section 6.1, while you outline theoretical and practical contributions, I find the explanation of managerial implications to be somewhat limited in terms of meaning. At present, these insights feel a bit left to chance rather than being the result of a rigorous analysis of the model's output. I suggest you go into more detail by explaining how a manager should practically interpret the SHAP results to change or support their stock policies. Without this depth, the "practical contribution" remains more of a statement of intent than a concrete tool for decision-makers.

Thank you for the opportunity to review this interesting work. Given the structural and depth-related issues identified, I believe the manuscript requires significant refinement before it can be considered for publication.

Author Response

Please, consult the attached file. Thank you.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors,

Kindly check for the English correction, otherwise it looks good.

Comments on the Quality of English Language

The English could be improved to more clearly express the research.