A New Phase Classifier with an Optimized Feature Set in ML-Based Phase Prediction of High-Entropy Alloys
Round 1
Reviewer 1 Report
In this paper, the authors have used machine learning methods for modelling the prediction of alloy phases. The problem of data class imbalance frequently encountered in data-driven modelling with alloy phase data was also researched, and combined EL, ADASYN oversampling methods with cost-sensitive learning mitigated the problem. In the end, the classifier achieved an accuracy of 95.44 %. The authors use machine learning to solve the problem of alloy phase classification in materials design, which is very interesting. However, the following issues still need to be addressed before publication:
1. The format of the features in Fig.2 and 3 should be uniform with the main text, e.g. “δTm”, “L7”, etc. should be consistent with “δTm”, “L7” in the main text.
2. The feature selection methods described by the authors in "3.2. Feature set" can be summarized.
3. In the Introduction or Methodology section, a large number of abbreviations are mentioned. It is suggested that abbreviations that occur infrequently be deleted.
Author Response
Please see the attachment
Author Response File: 
Author Response.docx
Reviewer 2 Report
In this work, the authors build a machine learning-based classifier for the phase prediction of HEAs. However, this work lacks innovation and convincingness. The results of this work do not exceed or even reach the level of previous work. Therefore, I can not recommend this manuscript to publish.
Comments：
1. Phase classification is one of the important scientific problems of HEAs, and a lot of research work has been done. In the Introduction, the summary of previous work in machine learning-based classification is not sufficient. Some influential articles have not been summarized and reviewed, for example:
1) Machaka, R. Machine learning-based prediction of phases in high-entropy alloys. Comp Mater Sci 188 (2021)
2) Zhang, Y. et al. Phase prediction in high entropy alloys with a rational selection of materials descriptors and machine learning models. Acta Mater 185, 528-539 (2020).
3) Mandal, P., Choudhury, A., Mallick, A. B. & Ghosh, M. Phase Prediction in High Entropy Alloys by Various Machine Learning Modules Using Thermodynamic and Configurational Parameters. Met Mater Int 29, 38-52 (2023).
2. In section 2.2, A lot of words are devoted to introduction of various algorithms, which is completely unnecessary.
3. The oversampling methods was used to solve the imbalance issue of dataset. How did the authors avoid overfitting problems?
4. The 10-fold method was used to evaluate the algorithm accuracy. It is not enough to judge the generalization ability of the model. The unseen data should be used.
5. The interpretability of the model is not discussed at all.
The English could be improved.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 3 Report
It is interesting work. Introduction is well written and results and discussion is proper. I recommend this article for publication.
However, some minor revisions and grammar mistakes should be solved.
(1) Please comment on the accuracy of results by adopted model.
(2) Validation of theoretical model should be addressed.
(3) How did you confirm that adopted alloy is high entropy alloy. have you calculated the entropy values?
(4) Grammer and English language should be improved to further enhance the quality of manuscript.
English and grammar should be improved.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
In the revision, the authors responded the comments from reviewers one by one. The responses are reasonable. According to the comments, the manuscript has been carefully revised and the quality of it is significantly improved. Therefore, I think this version can be recommend to publish.
Reviewer 3 Report
Thank you for incorporating the changes according to my comments.
