Comparative Analysis of Machine Learning Models for Predicting Viscosity in Tri-n-Butyl Phosphate Mixtures Using Experimental Data

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study proposes utilizing Machine Learning techniques to automate TBP mixture viscosity prediction, evaluating five algorithms.  The findings in the work underscore the transformative potential of machine learning in optimizing hydrometallurgical processes involving TBP mixtures, promising enhanced efficiency and precision while mitigating operational risks. 

The work is novel and interesting to the readers, and the presentation is good as well. But still it can be improved as follow before publication.

1. The abstract can be improved for fluent reading.

2. The introduction can be improved as well.

3. The figures in the paper must be improved, e.g. Figure 2 should be aligned center, and the size of all of the figures should be uniformed. Also the symbol of (a) (b) should also be uniformed, and the resolution of  some  figures are too low.

4. The conclusions are too simple and the main results should be displayed in the conclusions. 

Comments on the Quality of English Language

Minor editing of English language required.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

This study proposes utilizing Machine Learning techniques to automate TBP mixture viscosity prediction, evaluating five algorithms. The results show that the neural network (NN) model has the most accurate prediction results. The idea of the article is clear and the argument is sufficient, but the research method is conventional, so the innovation is not strong. The following are some suggestions to improve the article:

 

More serious problems:

 

1. The process in Figure 1 is too concise and does not reflect the research content of model comparison, and the text in the box does not correspond to the title of each section. It is suggested to modify the flow chart according to the title of each section and reflect the focus of the article on "model comparison"

 

2. On line 86, what is the "9" in "4599 = 511×9". This is the first time it has appeared and is not described in the article.

 

3. On line 114, the expression of "moderate correlations" is not scientific. The strength of relevance should be illustrated by numerical examples.

 

4. On line 124, you don't explain clearly why you want to explore the relationship between viscosity, temperature and TBP.

 

Detailed points in the article:

 

1. Without introducing what R2 is in the abstract.

 

2. Variables in the formula should be italicized.

 

3. Error indent on line 274.

 

Comments on the Quality of English Language

n

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

Dear authors, 

the manuscript is interesting. However, I have some comments and suggestions which could improve the manuscripts quality. 

1. In the introduction please emphasize the novelty and the idea of this paper as much as possible. Add the hypotheses in form of questions. At the end of the Intro section please add the short description of each of the following manuscript sections. 

2. In graphs such as figure 1 and others use the same font as in text (TimesNewRoman). All plots in the manuscript were done using the matplotlib. Here is the code for adjusting the font size and set the font style to TimesNewRoman. 

import matplotlib.pyplot as plt SMALL_SIZE = 20
MEDIUM_SIZE = 22
BIGGER_SIZE = 24
 plt.rc('font', size=SMALL_SIZE) # controls default text sizes plt.rc('axes', titlesize=SMALL_SIZE) # fontsize of the axes title plt.rc('axes', labelsize=MEDIUM_SIZE) # fontsize of the x and y labels plt.rc('xtick', labelsize=SMALL_SIZE) # fontsize of the tick labels plt.rc('ytick', labelsize=SMALL_SIZE) # fontsize of the tick labels plt.rc('legend', fontsize=SMALL_SIZE) # legend fontsize plt.rc('figure', titlesize=BIGGER_SIZE) # fontsize of the figure title

plt.rcParams["font.family"] = "Times New Roman"

Add the grids to all plots in the manuscript this is mandatory. 

4. Correlation matrix was done using matplotlib and seaborn library. Please add numbers to the correlation matrix heatmap. Here is the code. 

import matploitlib.pyplot as plt 

import seaborn as sns

plt.figure(figsize=(12,8))

sns.heatmap(data.corr(), annot=True)

plt.show()

 

5. You have used R^2, MAE and RMSE, You should consider using MAPE, RMSE and kling-gupta efficiency. For kling gupta efficiency please check the paper.   On the application of symbolic regression in the energy sector: Estimation of combined cycle power plant electrical power output using genetic programming algorithm

 

6. The discussion is too short please create a detailed discussion section. 

7. The Conclusion section must be restructured. First paragraph - short description of what was done in the manuscript. The second paragraph - answers the hypotheses defined in the intro section based on the discussion section. Third paragraph are pros and cons of the proposed research methodology. Pros and cons are Advantages and disadvantages. The final paragraph is directions for future work i.e. improvements that are based on the disadvantages of the proposed research methodology. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3. 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

Comments and Suggestions for Authors

This manuscript introduces Machine Learning techniques to predict the viscosity of TBP mixtures. After selecting features by PCA, the author tested five different ML models and compared their results. The followings are my comments and suggestions.

 

1. In Figure 2 (i), it seems the black line should be the normal distribution, not the blue line shown in legend.

 

2.     In Figure 2, it is confusing that the bar chart has different width. Can the author explain more about it?

 

3.     The results of PCA suggests only reducing the dimension by 1, which is not a significant reduction. Did the author test the error of models if they chose a lower tolerance for PCA and reduced more dimensions?

 

4.     Logistic Regression (LR) is commonly used for classification problem, not regression problem. In this manuscript, using LR for a regression problem is equivalent to a single layer NN with sigmoid activation function. Can author explain why they want to introduce LR in this problem?

 

5.     In addition to LR, the errors for other 4 models are very close (R^2 are all larger than 99%). From my understanding, they are all accurate enough. Does such small discrepancy support their statement that NN model is the best one? Because I think these differences can be just caused by the selection of some parameters in training.

 

6.     I suggest adding the comparisons of computational cost (training and inference) to provide readers a more comprehensive understanding about the ML models in the predictions of viscosity in TBP mixtures.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Dear authors, 

the manuscript has been modified and improved according to my suggestions and comments. The manuscript can be accepted in this form. 

 

Reviewer 4 Report

Comments and Suggestions for Authors

The author answered my questions and revised the manuscript accordingly. I suggest accepting the paper.