Machine Learning Prediction of Critical Temperature of Organic Refrigerants by Molecular Topology
Round 1
Reviewer 1 Report
The present paper "Machine learning prediction of critical temp..." is a paper about classifying and predicting the critical temperature of organic liquids from their structure input data. The prediction of drug molecules has been quite difficult by machine learning with much lower scores but the present case is much more limited in the sense that the database of refrigerants is limited in variation. Other similar papers with the same goal of prediction is cited in references 10, 30 and published 5 years earlier in the article Su, W. et al Chem Industry 67 (11) (2016).
The paper is well written and deserves to be published.
Author Response
1: The present paper "Machine learning prediction of critical temp..." is a paper about classifying and predicting the critical temperature of organic liquids from their structure input data. The prediction of drug molecules has been quite difficult by machine learning with much lower scores but the present case is much more limited in the sense that the database of refrigerants is limited in variation. Other similar papers with the same goal of prediction is cited in references 10, 30 and published 5 years earlier in the article Su, W. et al Chem Industry 67 (11) (2016).
The paper is well written and deserves to be published.
Response: Thank you for your positive evaluation of our work. We are very grateful for your direct recommendation of publication for our manuscript.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The article “Machine learning prediction of critical temperature of organic refrigerants by molecular topology ” is publishable after major revision.
The authors proposed a machine learning prediction of critical temperature of organic refrigerants by molecular topology. However, their results are lacking physical insight. In certain cases, it has been noticed that some model did not perform well, but the reason behind that was unexplained. Therefore, I am not recommending it for the publication in processes journal in its current form.
Author Response
1. The article “Machine learning prediction of critical temperature of organic refrigerants by molecular topology ” is publishable after major revision.
The authors proposed a machine learning prediction of critical temperature of organic refrigerants by molecular topology. However, their results are lacking physical insight. In certain cases, it has been noticed that some model did not perform well, but the reason behind that was unexplained. Therefore, I am not recommending it for the publication in processes journal in its current form.
Response: We appreciate the reviewer’s valuable comment very much. We fully understand the reviewer’s concerns. And we apologize for the lack of completeness and clarity of the expressions in the article. As suggested by reviewer, we have added a more detailed interpretation regarding the situation that some models did not perform well on page 5-6.
The application of machine learning on physical property prediction can be divided into two steps. Firstly, a molecular structure is converted into a vector of features, secondly the algorithms map between feature vectors and the property. In this work, molecular fingerprints and topological indices are used as features to predict the critical temperature. In this process, the representation ability of fingerprints for structures and the fitting ability of the algorithms will both affect the model performance. Therefore, by analyzing the mechanism of fingerprints, and the problems encountered frequently in the modeling process, we explained the phenomenon that some models have limited performance and gave some suggestions for establishing better models.
Author Response File: Author Response.pdf
Reviewer 3 Report
The article investigates the Tc of a group of organic refrigerants using engineering designs and machine learning algorithms. My comments are as follows:
- Although the article is well-written, there are minor glitches that can be overcome by proofreading.
- The introduction needs to be enhanced by adding some information about organic refrigerants for the readers to follow.
- An appendix is preferably added with all nomenclature.
- Only key result is to be added at the end of the abstract.
- Any special cases to your study?
- Punctuation is missing after some equations.
Accordingly, I advise that the manuscript is put under minor revision until the inquires addressed are responded to.
Author Response
The article investigates the Tc of a group of organic refrigerants using engineering designs and machine learning algorithms. My comments are as follows:
Although the article is well-written, there are minor glitches that can be overcome by proofreading.
1. The introduction needs to be enhanced by adding some information about organic refrigerants for the readers to follow.
Response: Thanks a lot for the reviewer. According to the reviewer’s advice, we have added the relationship among the organic refrigerants, working fluids and thermodynamic cycles in the first paragraph of introduction：
The propose of Carbon Neutrality will accelerate the utilization of renewable energy, such as solar energy, geothermal energy and so on[1, 2]. And the thermodynamic cycles, including the novel power systems represented by organic Rankine cycle (ORC), and re-frigeration/heat pump cycles represented by vapor compression cycle, are the effective ap-proaches to use these medium and low energy. The working fluid is the energy carrier of thermodynamic cycles, which plays a key role in designing and enhancing the thermo-dynamic cycles[3, 4]. Organic refrigerants, as a kind of compounds with low boiling point, apart from their application in refrigeration industry, also have unique advantage to be used as working fluids in ORC to recover low-grade energy and improve the energy utili-zation efficiency.
2. An appendix is preferably added with all nomenclature.
Response: We appreciate the reviewer’s useful suggestion very much. We have added the nomenclature on page 10.
3. Only key result is to be added at the end of the abstract.
Response: Thanks a lot for the reviewer’s useful advice. To concisely introduce our work, the abstract has been rewritten as follows:
In this work, molecular structures, combining with machine learning algorithms, were applied to predict the critical temperatures (Tc) of a group of organic refrigerants. Aiming at solving the problem that previous models cannot distinguish isomers, a topological index was introduced. The results indicate that the novel molecular descriptor ‘molecular fingerprint + topological index’ can effectively differentiate isomers. The average absolute average deviation between the predicted and experimental values is 3.99%, which proves reasonable prediction ability of the present method. In addition, the performance of the proposed model was compared with that of other previously reported methods. The results show that the present model is superior to other approaches, with respect to accuracy.
4. Any special cases to your study?
Response: Thanks a lot for the reviewer. This work collected the critical temperature experimental data of alkanes, halogenated alkanes, alkenes, halogenated alkenes and ethers. But alcohols and amines can also be applied as working fluids in thermodynamic cycles. These two types of chemicals are lack of experimental values. Therefore, they are neglected in order to prevent that too small molecules in the fitting process reduce the accuracy of models. In the future with more reliable experimental points enriching the database, a more comprehensive prediction model can be established by using the method in this paper.
5. Punctuation is missing after some equations.
Response: Thank you so much for your careful check. We have carefully checked the manuscript and added the punctuation after these equations.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The manuscript “Machine learning prediction of critical temperature of organic refrigerants by molecular topology" is publishable in the Processes journal in its current form.
