Review Reports - Assessing the Performance of Machine Learning Algorithms for Soil Classification Using Cone Penetration Test Data

Round 1

Reviewer 1 Report

This manuscript describes the efficiency of various machine learning (ML) algorithms in classifying soils based on Robertson's soil behavioral types. The study employed four ML algorithms, including Artificial Neural Network (ANN), Logistic Regression (LR), Support Vector Machine (SVM), and Gradient Boosting Machine (GBM), to classify soils based on cone penetration (CPT) datasets. So far, the paper includes new contributions. However, a number of issues in the manuscript are expected to be solved. Besides, there are some suggestions and questions that should be addressed.

-The current structure of the introduction is not well organized and long. Additionally, the authors need to be improved the last part of the introduction considering the main theme/objectives and findings of the study.

- It will be more appropriate if you add a paragraph at the end of the introduction section illustrating the layout of the paper.

-Recommendations for practice and prospective work are missing.

-Authors need to revise and improve based on key findings with clearer summarized information and research significance.

-In Eq. 1 Pa is not introduced.

-The figures should be increasing in the order of occurrence. E.g. in section 2 first, Fig. 1 is supposed to be introduced.

In the conclusion, please clearly mention the practical outcome of the study. Please clarify what the employed technique offers to the engineering field compared to other methods available in the literature.

Author Response

Dear Reviewer,

We are writing to respond to the comments you provided on our manuscript, "Assessing the Performance of Machine Learning Algorithms for Soil Classification Using Cone Penetration Test Data," which we submitted to the Journal of Applied Sciences for consideration.

Firstly, we would like to express our gratitude for taking the time to review our work and provide valuable feedback. Your insights have greatly helped us refine and improve our research, and we appreciate your careful consideration of the manuscript.

In response to your comments, we have made modifications to our revised manuscript. We believe that these revisions have greatly strengthened the manuscript and addressed the concerns you raised. We are confident that the manuscript now meets the standards for publication in the Journal of Applied Sciences. Please see the attachment.

Once again, we would like to express our appreciation for your time and effort in reviewing our manuscript. Your feedback has been invaluable in improving the quality of our work.

Thank you for your consideration.

Sincerely,

Chala Ayele and Prof. Richard Ray

Author Response File: Author Response.pdf

Reviewer 2 Report

It has certain scientific value.

This kind of research is very common.

The innovation of the paper is not too high.

The detailed comments are in the attachment.

Authors need to be aware that more references need to be added.

The author carefully revised and responded all the questions and comments.

Comments for author File: Comments.docx

Author Response

Dear Reviewer,

Once again, we would like to express our appreciation for your time and effort in reviewing our manuscript. Your feedback has been invaluable in improving the quality of our work.

Thank you for your consideration.

Sincerely,

Chala Ayele and Prof. Richard Ray

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper applies four machine learning algorithms on classifying soil types. Different ML methods are well interpreted and the prediction results are compared and analyzed. Overall, the paper is well organized and its science is scholarly developed. I have several comments below:

1. It is well known that there are many machine learning methods, such as CNN, KNN, LVQ, etc. Why do you choose these four for study?

2.On lines 213, 231, 284, 285 and in table 11, you mentioned overall accuracy. Please explain how did you derive it, by averaging the accuracy of different parameters?

3.In sections 4.2, 4.3, and 4.4, you just briefly explained the principles of the ML models (different models have different parameters), but normally when a ML model is adopted, it is better to clarify what kind of parameters are used in the application.

4.The conclusion is too tedious, it is better to make it more concise. For example, the sentence "Conventional methods for soil classification... ...laboratory test" in the first paragraph is unnecessary, and the second paragraph may be better to move to front sections.

5.On line 63, you mentioned Figure 2 first, and then on line 73, you mentioned Figure 1. This order is reversed. The figures should be mentioned in order.

6.In Figure 2, the legends are missing, otherwise we don't know what the different color lines mean.

7.On line 102, delete "on"

8.On line 321, change "we" to "was"

9.It is better to address the significance of your work in the conclusion section.

10. One important task for ML learning is preparing labels, in this work you just used Roberton's classification. It may be better to provide more explanations regarding why his parameters are representative and are good labels.

Author Response

Dear Reviewer,

Once again, we would like to express our appreciation for your time and effort in reviewing our manuscript. Your feedback has been invaluable in improving the quality of our work.

Thank you for your consideration.

Sincerely,

Chala Ayele and Prof. Richard Ray

Author Response File: Author Response.pdf

Reviewer 4 Report

This paper performs soil classification using various machine-learning methods. The subject is interesting, but the following comments are suggested to consider.

1. Why did the authors choose the four machine learning methods? The authors can also consider other commonly used methods such as LSTM and CNN.

2. How to determine the hyperparameters of the selected machine learning methods? Bayesian optimization and/or random optimization methods are recommended. More details can refer to the Probabilistic framework with Bayesian optimization for predicting typhoon-induced dynamic responses of a long-span bridge, and this reference should be added.

3. Data quality is critical for data-driven machine learning methods, and anomalies usually exist in the monitoring data. Thus, the data pre-processing is essential for such methods, and the anomaly detection methods, such as Anomaly detection of structural health monitoring data using the maximum likelihood estimation-based Bayesian dynamic linear model, should be briefly reviewed in the Introduction.

Author Response

Dear Reviewer,

Once again, we would like to express our appreciation for your time and effort in reviewing our manuscript. Your feedback has been invaluable in improving the quality of our work.

Thank you for your consideration.

Sincerely,

Chala Ayele and Prof. Richard Ray

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I suggest accepting the paper.

Author Response

Dear reviewer

Thank you so much for your time and valuable comments.

Reviewer 2 Report

The author carefully revised and responded all the questions and comments.

Although there are still some problems in the author's revision.

It is recommended to accept and publish.

Author Response

Dear reviewer

Thank you so much for your time and valuable comments