Tree-Based Machine Learning and Nelder–Mead Optimization for Optimized Cr(VI) Removal with Indian Gooseberry Seed Powder
, D. Krishna 2,†, Upendra Kumar Potnuru 3,†, Manohar Mishra 4,*, Salman S. Alharthi 5,* and Ravindranadh Koutavarapu 6,†Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsTitle: Tree-Based Machine Learning and Nelder-Mead Optimization for Optimized Cr(VI) Removal by Indian Gooseberry Seed Powder
Manuscript ID: water-3052059
Authors: Lakshmana Rao Kalabarige, D. Krishna, Upendra Kumar Potnuru, and Manohar Mishra
Comments to the corresponding author:
This article contributes to the field on an important topic of water treatment. In this work, Indian Gooseberry Seed Powder was used for the sequestration of Cr(VI) from aqueous solution. As I read this paper, I found that the paper has some issues that prevent it from being published in its current state. The manuscript needs to be revised extensively before it can be published. As such, the authors should seriously consider all of the following comments:
1. The paper should be checked for subscripts and superscripts.
2. Avoid unnecessary paragraphs.
3. Many acronyms are introduced directly. First give the full form and then follow the acronym throughout the paper.
4. Give a novelty statement.
5. For Cr(VI) adsorption on plant, here are the references to cite: Removal of Cr (VI) from aqueous solution on seeds of Artimisia absinthium (novel plant material); Removal of Cr(VI) from electroplating wastewater using fruit peel of Leechi (Litchi chinensis).
6. For BBD and ANN, here is a reference to cite: The artificial neural network and Box-Behnken design for Cu2+ removal by the pottery sludge from water samples: Equilibrium, kinetic and thermodynamic studies.
7. Section 2.1. “desired particle sizes”. Mention the number.
8. As the experimental study was conducted, optimization should be used to confirm its results. Hence, the results of experimental studies should also be presented and discussed.
9. The study seems incomplete without characterization analysis.
10. Rewrite conclusion. It is similar to the abstract.
Comments on the Quality of English Language
The paper should be checked for subscripts and superscripts.
Author Response
Response to Reviewer-1
This article contributes to the field on an important topic of water treatment. In this work, Indian Gooseberry Seed Powder was used for the sequestration of Cr(VI) from aqueous solution. As I read this paper, I found that the paper has some issues that prevent it from being published in its current state. The manuscript needs to be revised extensively before it can be published. As such, the authors should seriously consider all of the following comments:
Author Response: The authors would like to thank the Editor and Reviewers for the time they spent reviewing our manuscript. The suggestions provided were constructive to enhance the quality of the manuscript. We have tried to answer all your questions/remarks/suggestions to the best of our knowledge. We hope you find the answers satisfactory and the manuscript of better quality.
Reviewer #1 Concern #1: The paper should be checked for subscripts and superscripts.
Author Response: Thank you for your suggestions. As suggested by the reviewer, we have checked and corrected all typo errors in subscripts and superscripts.
Reviewer #1 Concern #2: Avoid unnecessary paragraphs.
Response: Thank you for your suggestions. As suggested by reviewer, we have reduced unnecessary paragraphs and same is reflected in revised manuscript.
Reviewer #1 Concern #3: Many acronyms are introduced directly. First give the full form and then follow the acronym throughout the paper.
Response: Thank you for your suggestions. As suggested by the reviewer, we have checked all the acronyms used in this manuscript and corrected wherever necessary and reflected in revised manuscript.
Reviewer #1 Concern #4: Give a novelty statement.
Response: Thank you for your suggestions. The novelty of this work is presented in the revised paper in the introduction section of this manuscript.
The introduction section sub-divided as Motivation section, Literature review section, Research gap section, Major contribution section; and organisation section along with novelty section. Hence, it was highlighted with blue color in the revised manuscript.
Reviewer #1 Concern #5: For Cr(VI) adsorption on plant, here are the references to cite: Removal of Cr (VI) from aqueous solution on seeds of Artimisia absinthium (novel plant material); Removal of Cr(VI) from electroplating wastewater using fruit peel of Leechi (Litchi chinensis).
Response: As suggested by reviewer, we felt that the suggested references were appropriate and cited in revised manuscript with reference numbers 29 and 30.
Reviewer #1 Concern #6: For BBD and ANN, here is a reference to cite: The artificial neural network and Box-Behnken design for Cu2+ removal by the pottery sludge from water samples: Equilibrium, kinetic and thermodynamic studies.
Response: As suggested by reviewer, we felt that the suggested reference was appropriate and cited in revised manuscript with reference number 23.
Reviewer #1 Concern #7: Section 2.1. “desired particle sizes”. Mention the number.
Response: As suggested by the reviewer, we have included size of particles and updated in revised manuscript.
Author Action: Desired particle sizes of 63 µm, 89 µm, and 125 µm.
Reviewer #1 Concern #8: As the experimental study was conducted, optimization should be used to confirm its results. Hence, the results of experimental studies should also be presented and discussed.
Response:
The suggestion from the reviewer to utilize optimization for validating the experimental results and to thoroughly present and discuss these findings is highly valued. We concur that optimization is essential for confirming the experimental outcomes and guaranteeing the reliability of our results. Consequently, we have updated our manuscript to incorporate a detailed analysis of the optimization results and their correlation with our experimental data.
Revisions Made:
The results obtained from ETR-Nelder-Mead optimization were validated through experimentation as reported in Table 7. The results shows that the experimental outcome for optimal parameters such as optimal-Initial-concentration of Cr(VI) with 99.25, optimal pH with 4.97, and optimal Adsorbent dosage (g/L) (9.62) strongly aligns with experimentation results with an 6.56% of error rate.
|
Optimal Initial concentration of Cr(VI) |
Optimal pH |
Optimal Adsorbent dosage (g/L) |
Optimal Cr(VI) removal % |
Obtained Cr(VI) removal % |
% Error |
|
99.25 |
4.97 |
9.62 |
85.11 |
79.75 |
6.72 |
Reviewer #1 Concern #9: The study seems incomplete without characterization analysis.
Response:
We appreciate your insightful feedback and the emphasis on the significance of characterisation analysis in our research. We recognize that characterisation analysis is essential for comprehending the properties and behaviour of the materials utilized.
In our present research, our main emphasis was on improving the removal efficiency of Cr(VI) by utilizing tree-based machine learning models and Nelder-Mead optimization. Our main goal was to assess the effectiveness and enhancement of the elimination process. Nevertheless, we recognize that conducting a comprehensive analysis of the Indian Gooseberry Seed Powder could offer more profound understanding of its composition, functional groups, and adsorption mechanisms.
In our future work, we intend to conduct thorough analysis of the characteristics, which will entail utilizing methods like
- Fourier Transform Infrared Spectroscopy (FTIR): To identify the functional groups present in IGSP.
- Scanning Electron Microscopy (SEM): To observe the surface morphology and porosity of IGSP.
- X-Ray Diffraction (XRD): To determine the crystalline structure of IGSP.
Reviewer #1 Concern #10: Rewrite conclusion. It is similar to the abstract.
Response: As suggested by the reviewer, we agreed with the perception of reviewer, and we have updated the previous conclusion as follows.
The previous conclusion:
A detailed batch experimental study was carried out for the removal of chromium (VI) from wastewater by Indian gooseberry seed powder. The objective of the present study was to find out the optimum process parameters, using ML models and Nelder-Mead optimization for the removal of Cr(VI) from wastewater by Indian gooseberry seed powder as adsorbent.
Among all the models, DTR, RFR, and ETR displayed exceptional performance, achieving -scores of 99\%, respectively. Remarkably, all models exhibited an RRMSE of 1\%. These statistics further support the notion that these models performed exceptionally well.
Notably, the ETR approach outperformed all other models, showcasing a remarkable $R^2$-score. Furthermore, the MAE, MSE, and RMSE of the ETR indicated significantly lower error rates compared to the other models. Moreover, the percentage of Cr(VI) removal for Initial concentration of Cr(VI) of 99.55 mg/L, pH of 4.0, and Adsorbent dosage of 8.0 (g/L) gives highest Cr(VI) removal percentage such as 78.21\% and 78.107 respectively by DTR and RFR is as shown in Table 6. The removal efficiency of ETR model was 85.11\% for Initial concentration of Cr(VI) of 99.24284408 mg/L, pH of 4.96268861, and Adsorbent dosage of 9.61957019 (g/L). Furthermore, the optimization results presented in Table 6 reported optimized values obtained by ML based Nelder-Mead optimization approach to obtain maximum Cr(VI) removal percentage. That is, the DTR-Nelder-Mead, RFR-Nelder-Mead and ETR-Nelder-Mead optimization approaches were able to obtain maximum of 78.107\% to 85.11\% of Cr(VI) removal which reported increase in Cr(VI) removal percentage ranging from 4.66\% to 11.56\% than the Cr(VI) removal percentage of 73.55\% obtained by experimentation for Initial concentration of Cr(VI) of 20 mg/L, pH of 2.0, and Adsorbent dosage of 8.0 (g/L).
In summary, based on the comprehensive evaluation, it is concluded that the ML models exhibited strong performance overall. ETR, in particular, stood out as the top performer, demonstrating superior accuracy with minimal error rate and obtains higher Cr(VI) removal percentage than all other models. Furthermore, The ML based Nelder-Mead optimization results indicated that the maximum of 85.11\% chromium VI removal is possible with Initial concentration of Cr(VI) of 99.24284408 mg/L, pH of 4.96268861, and Adsorbent dosage of 9.61957019 (g/L). These results indicated that the maximum Cr(VI) removal percentage by ML modeling and ML based Nelder-Mead optimization reported 11.56\% of increase in Cr(VI) removal percentage for different absorptions which were not tested during experimentation.
The Revised Conclusion:
This study examined a novel approach for percentage of Cr(VI) removal efficiency from wastewater by employing ML models combined with Nelder-Mead optimization. Especially, this combined strategy for adsorption capacity estimation using any solute or material has not been previously reported in the literature. This research not only leverages the power of ML for accurate prediction of percentage of Cr(VI) removal efficiency but also utilizes Nelder-Mead optimization to identify optimal process parameters which maximize Cr(VI) removal efficiency, ultimately achieving a significant enhancement in Cr(VI) removal efficiency compared to standalone experimentation.
In this work, particularly the ETR, achieved outstanding performance in predicting Cr(VI) removal. The ETR model exhibited a remarkable -score of 99% and significantly lower error rates compared to other models. Furthermore, ML-based Nelder-Mead optimization identified optimal process parameters, leading to a maximum Cr(VI) removal efficiency of 85.11%. This represents an improvement of up to 11.56% compared to standalone experimentation.
These results demonstrate the effectiveness of the combined ML and optimization approach for maximizing Cr(VI) removal using Indian gooseberry seed powder. This research work paves the way for further exploration of ML-driven strategies in wastewater treatment applications.
Thank you again for reviewing our paper and providing valuable suggestions to improve the quality of the paper.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe manuscript "Tree-Based Machine Learning and Nelder-Mead Optimization for Optimized Cr(VI) Removal by Indian Gooseberry Seed Powder" by L. R. Kalabarige describes a series of statistical/Machine Learning methods to predict the removal of Cr(VI) from wastewater using the crushed powder of an indian seed. The work may be of interest for people working in the field, though it is probably a bit too specialized (but I guess Waters audience is quite specialized as well.
In some parts of the paper the thread is difficult to follow and should be made clearer. The first question regards the necessity of building a "synthesized database". The experimental dataset shown (Table 1) contains 56*4 = 224 points (=compound + properties, like pH etc) is then converted into a larger one (2000*4) by some kind of extrapolation/generation, that is used for model training.
Excessive use of digits (e. g. 99.24284408 mg/L, pH of 4.96268861), is very frequent in the text. They are certainly unphysical. Also numerical descriptors (like R2) are given with variable numbers of digits.
"...ML modelling and ML based Nelder-Mead optimization reported 11.56% of increase in Cr(VI) removal percentage than the experimentation results for different absorptions which are not tested during experimentation. , tree-based regression models have not been previously employed. The part after THAN is quite unreadable,
Page 4: "than the experimentation results for different absorptions which are not tested during experimentation. , tree-based regression models have not been previously employed."
Please enlarge equation 1
Obscure writing:
Page 6
"This conversion process not bounded to certain range."
"Finally, tested each model with testing data."
Comments on the Quality of English LanguageThe language is quite fine
Author Response
Response to Reviewer-2:
The manuscript "Tree-Based Machine Learning and Nelder-Mead Optimization for Optimized Cr(VI) Removal by Indian Gooseberry Seed Powder" by L. R. Kalabarige describes a series of statistical/Machine Learning methods to predict the removal of Cr(VI) from wastewater using the crushed powder of an indian seed. The work may be of interest for people working in the field, though it is probably a bit too specialized (but I guess Waters audience is quite specialized as well.
Author Response: The authors would like to thank the Editor and Reviewers for the time they spent reviewing our manuscript. The suggestions provided were constructive to enhance the quality of the manuscript. We have tried to answer all your questions/remarks/suggestions to the best of our knowledge. We hope you find the answers satisfactory and the manuscript of better quality.
Reviewer #2 Concern #1: In some parts of the paper the thread is difficult to follow and should be made clearer. The first question regards the necessity of building a "synthesized database". The experimental dataset shown (Table 1) contains 56*4 = 224 points (=compound + properties, like pH etc) is then converted into a larger one (2000*4) by some kind of extrapolation/generation, that is used for model training.
Response: We are thankful to the reviewer for raising a valid point regarding the necessity of a synthesized dataset when we have an initial dataset of 55 data points.
The experimental data provides valuable insights with 55 instances. However, it has limitations for training robust ML models. The substantial amount of data is required for the effective training of ML models to understand hidden patterns, and complex non-linear relationships between features. Hence, it is difficult to have an effective training of ML models to capture complex patterns, understand full range of interactions between the variables, and achieve optimal model performance. Furthermore, a larger dataset with 2000 instances, allows the model to learn from a broader range of scenarios. This enhances the model's generalizability, meaning it can perform well on unseen data not included in the training set. Moreover, the randomization approach employed to built the synthesized dataset expands the range of feature values beyond those observed in the experiments (Table 1). This enables the exploration of a wider parameter space, potentially leading to the identification of optimal conditions that might not have been captured in the original experimental design.
Reviewer #2 Concern #2: Excessive use of digits (e. g. 99.24284408 mg/L, pH of 4.96268861), is very frequent in the text. They are certainly unphysical. Also, numerical descriptors (like R2) are given with variable numbers of digits.
Response: As reviewer suggested, the excessive and variable length digits after decimal is reduced and made uniform.
Reviewer #2 Concern #3: "...ML modelling and ML based Nelder-Mead optimization reported 11.56% of increase in Cr(VI) removal percentage than the experimentation results for different absorptions which are not tested during experimentation. , tree-based regression models have not been previously employed. The part after THAN is quite unreadable,
Response: As suggested by reviewer, the sentence framing is not appropriate and place for the below paragraphs is not appropriate. Hence, we have corrected the entire paragraph and removed from introduction section and added in result section by removing duplication in writeup.
The removed paragraphs from introduction section are as follows:
Notably, the ETR approach outperformed all other models, showcasing a remarkable $R^2$ score. Furthermore, the MAE, MSE, and RMSE of the ETR indicated significantly lower error rates compared to the other models. Moreover, the percentage of Chromium (VI) removal for Initial concentration of Cr(VI) of 99.55 mg/L, pH of 4.0, and Adsorbent dosage of 8.0(g/L) gives highest Cr(VI) removal percentage such as 78.21\% and 78.107 respectively by DTR and RFR is as shown in Table 6. The removal efficiency of ETR model was 85.11\% for Initial concentration of Cr(VI) of 99.24284408 mg/L, pH of 4.96268861, and Adsorbent dosage of 9.61957019(g/L). Furthermore, the optimization results presented in Table~\ref{tab6} reported optimized values obtained by ML based Nelder-Mead optimization approach which obtain maximum Cr(VI) removal percentage. That is, the DTR-Nelder-Mead, RFR-Nelder-Mead and ETR-Nelder-Mead optimization approaches were able to obtain maximum of 78.107\% to 85.11\% of Cr(VI) removal which reported increase in Cr(VI) removal percentage ranging from 4.66\% to 11.56\% than the Cr(VI) removal percentage of 73.55\% obtained by experimentation for Initial concentration of Cr(VI) of 20 mg/L, pH of 2.0, and Adsorbent dosage of 8.0(g/L).
In summary, based on the comprehensive evaluation, it is concluded that the ML models exhibited strong performance overall. ETR, in particular, stood out as the top performer, demonstrating superior accuracy with minimal error rate and obtains higher Cr(VI) removal percentage than all other models. Furthermore, The ML based Nelder-Mead optimization results indicated that the maximum of 85.11\% chromium VI removal is possible with Initial concentration of Cr(VI) of 99.24284408 mg/L, pH of 4.96268861, and Adsorbent dosage of 9.61957019 (g/L). These results indicated that the maximum Cr(VI) removal percentage by ML modelling and ML based Nelder-Mead optimization reported 11.56\% of increase in Cr(VI) removal percentage than the experimentation results for different absorptions which are not tested during experimentation. Furthermore, tree-based regression models have not been previously employed. Hence, this study ventures to apply three different models and ML based optimization to analyze and compare their performance on experimental data.
The corrected paragraph is as follows and placed in result section:
Similarly, the DTR-Nelder-Mead, RFR-Nelder-Mead and ETR-Nelder-Mead optimization approaches were able to obtain maximum of 78.107\% to 85.11\% of Cr(VI) removal which reported increase in Cr(VI) removal percentage ranging from 4.66\% to 11.56\% than the Cr(VI) removal percentage of 73.55\% obtained by experimentation for Initial concentration of Cr(VI) of 20 mg/L, pH of 2.0, and Adsorbent dosage of 8.0(g/L). In summary, based on the comprehensive evaluation, it is concluded that the ML models exhibited strong performance overall. ETR and ETR-Nelder-Mead optimization, in particular, stood out as the top performer, demonstrating superior accuracy with minimal error rate and obtains higher Cr(VI) removal percentage than all other models.
Reviewer #2 Concern #4: Page 4: "than the experimentation results for different absorptions which are not tested during experimentation. , tree-based regression models have not been previously employed."
Response: This matter was corrected and removed from introduction section and kept in result section as mentioned in response to concern3.
Reviewer #2 Concern #5: Please enlarge equation 1
Response: As suggested, it was rewritten as follows and enlarged the equation using its acronyms. The same is highlighted in manuscript with blue color.
Percentage removal of Cr(VI) = ((a ∗ IC) + (b ∗ pH) + (c ∗ AD))
Reviewer #2 Concern #6: Obscure writing:
Page 6
“This conversion process not bounded to certain range.”
Response: As suggested by reviewer, it was identified that the above statement “This conversion process not bounded to certain range” doesn’t carry any meaning. Hence, it was removed from the manuscript.
"Finally, tested each model with testing data."
Response: The following paragraph under section 3.0.4 was modified by removing “Finally, tested each model with testing data." The revised manuscript highlighted the modified content in blue color.
The old write-up:
The proposed work applied tree-based regression models (as reported in Figure~\ref{fig1}) on synthesized dataset generated through curve-fitting on experimental data to build and train an ML model. Finally, tested each model with testing data. This work divided synthesized data into two portions such as training (85\%) and testing (15\%).
The modified writeup:
The proposed work employed tree-based regression models (as reported in Figure~\ref{fig1}) to attain higher prediction accuracy for Cr(VI) removal. The synthesized dataset was divided as training (85\%) and testing (15\%) sets.
Thank you again for reviewing our paper and providing valuable suggestions to improve the quality of the paper.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsTitle: Tree-Based Machine Learning and Nelder-Mead Optimization for Optimized Cr(VI) Removal by Indian Gooseberry Seed Powder
Manuscript ID: water-3052059
Authors: Lakshmana Rao Kalabarige, D. Krishna, Upendra Kumar Potnuru, and Manohar Mishra
Comments to the corresponding author:
While the authors have incorporated many changes, there are still a few important questions that need to be resolved before the manuscript can be published. Following are the comments:
1. As this is not a short communication, but a regular article, it is important to do an after adsorption analysis. After-adsorption characterizations are extremely important for improving the quality of the paper, matching it to the Water Journal, and providing more information about the adsorption process. I suggest that the authors perform SEM and FTIR.
2. Only the optimum values of various parameters, obtained through experimental processes, are included. It is important to discuss the results of these experiments.
Author Response
Response to Reviewer-1
Comments to the corresponding author:
While the authors have incorporated many changes, there are still a few important questions that need to be resolved before the manuscript can be published. Following are the comments:
Author Response: The authors would like to thank the Editor and Reviewers for the time they spent reviewing our manuscript. The suggestions provided were constructive to enhance the quality of the manuscript. We have tried to answer all your questions/remarks/suggestions to the best of our knowledge. We hope you find the answers satisfactory and the manuscript of better quality.
Reviewer #1
Comment 1: "As this is not a short communication, but a regular article, it is important to do an after adsorption analysis. After- adsorption characterizations are extremely important for improving the quality of the paper, matching it to the water journal, and providing more information about the adsorption process. I suggest that the authors perform SEM and FTIR."
Author Response:
We acknowledge the reviewer’s opinion. In order to give more thorough insights into the adsorption process, we appreciate the suggestion to include after adsorption characterizations. We performed the following analyses after taking this into account. The surface morphology of the Indian Gooseberry Seed Powder was examined using SEM analysis both before and after Cr(VI) adsorption as reported in Figure 5. The SEM images before and after Cr(VI) adsorption confirms that, the morphology seems to be irregular shaped sheet-like structure and didn’t find any significant changes after adsorption process. This confirms that the material exhibits same morphology before and after reaction. Thank you for your valuable comment, which enhances the quality of the manuscript.
We regret that we were not able to investigate the FT-IR measurements, which could definitely give us additional information. Unfortunately, we don’t have the facility of FT-IR instrument. We hope the reviewer understand the experimental deficiencies at the stage of the present experiments. We deeply appreciate the comment raised by the reviewer. Thank you very much.
Comment 2: “Only the optimum values for various parameters, obtained through the experimental process, are included. It is important to discuss the results of these experiments”.
Author Response:
We recognized the significance of offering a thorough analysis of the experimental findings in order to enhance comprehension of the optimization procedure and its consequences. We offered a thorough analysis of the experiment findings in the manuscript under section 4.1 as stated below and revised manuscript highlighted in blue color.
The results obtained from ETR-Nelder-Mead optimization were validated through experimentation as reported in Table 7 since strong prediction framework for the adsorption process was provided by the tree-based machine learning model, which further validated the results. The results shows that the experimental outcome for optimal parameters such as optimal-Initial-concentration of Cr(VI) with 99.25, optimal pH with 4.97, and optimal Adsorbent dosage (g/L) (9.62) strongly aligns with an 6.56% of error rate. In view of experimental results as well as model values, Indian gooseberry seed powder is an effective and low-cost adsorbent and suggested to use it for removal of hexavalent chromium from water as hexavalent chromium has the carcinogenic nature. Thank you for valuable comment.
___________________________________________________________________
Thank you again for reviewing our paper and providing valuable suggestions to improve the quality of the paper.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors have modified the manuscript according to my suggestions, therefore I think it can be considered for publication now.
Author Response
Response to Reviewer-2:
The authors have modified the manuscript according to my suggestions; therefore I think it can be considered for publication now.
Author Response: The authors would like to thank the Reviewers for the time they spent reviewing our manuscript and providing several constructive comments, and finally recommended for publication.
Author Response File: Author Response.pdf
