An Optimized Clustering Approach to Investigate the Main Features in Predicting the Punching Shear Capacity of Steel Fiber-Reinforced Concrete

Round 1

Reviewer 1 Report

This study introduces a novel system for solving engineering challenges based on the 19 features of the data. Due to the fact that data analysis involves various variations, the inclusion of common features can improve the performance and precision of models. Using a combination of optimization techniques (the K-means algorithm) and prediction techniques, this work presents a new system and method for identifying and analyzing similar and closely clustered data. The system created with the new sparrow search algorithm (SSA) has been revised as a hybrid solution to optimize development engineering issues. A series of laboratory experiments on samples of steel fiber-reinforced concrete yielded the information necessary for the proposed procedures (SFRC). To analyze the issue, the data were separated into distinct clusters based on shared characteristics. Each of the top clusters was then constructed using three predictive models, including multi-layer perceptron (MLP), support vector regression (SVR), and tree-based approaches, after the introduction of the clusters. This procedure will continue until the specified conditions are met. Accordingly, the K-means-artificial neural network 3 structure demonstrates the best accuracy and error performance.

Major: 

What is the rationale behind considering the proposed algorithms.? Please write the contribution using bullet format. 

Figure 12 requires detial desciption. It is difficult to understand. 

The future scope of this research is missing. Please include it. In addition, the label of y axis is missing. 

Compare this work with the state-of-the art. 

Include more recently published work in the review section. 

 

Author Response

Find attached file. 

Author Response File: Author Response.pdf

Reviewer 2 Report

The research topic is important and relevant. Identifying ideal clusters and applying new, more accurate and relevant clustering methods is an important research topic. 

The title is good overall, but should not be abbreviated as SFRC (steel fibre-reinforced concrete).

In the case of keywords, it is advisable to use relevant terms that are not included in the title.

The k-means is the most frequently applied method in the literature, however, due to its importance, the introduction section should be extended with an overview of clustering methods (hierarchical clustering, fuzzy clustering, density-based clustering, spectral clustering, etc.) and clustering validity indices, accuracy, error, etc. The introduction section should be structured.

At the end of the introduction, the objectives should be presented in separate bullet points for easy reference.

The sections on experimental setup and material and methods (K-means, Multi-Layer Perceptron (MLP), Random Tree (RT), Support Vector Regression (SVR), Sparrow search algorithm (SSA)) should be described and presented in a precise and sufficiently detailed manner. However, it would be important to present the optimal parameters: h, d, bc, fc, ?, ??.

The presentation of the steps of the K-means method is particularly important. Figure 3 is illustrative.

Why did they choose RMSE, MAE, and R2 criteria in their research? (It would be good to justify in the manuscript.)

The most important figure in the manuscript is Figure 4, which illustrates each step.

The labels on the x, y axis of the figure should be in larger font.

The results are illustrative, especially in Table 2. Results of various hybrid K-means models and Figure 12. Investigating the effect of input parameters on the structure of the selected K-means-403 ANN3 model.

The comparison of the results with other cross-generational results is missing in the manuscript and needs to be extended.

The conclusion section is adequate.

Author Response

Find the attached file. 

Author Response File: Author Response.pdf

Reviewer 3 Report

General comment:

This paper investigated the predicting methods of punching shear capacity of SFRC, the datasets were obtained based on the literature. There are three different models compared in terms of predicting accuracy.  Finally, the optimum parameters were determined based on the ANN models. There are some technique comments the author may consider for revising the manuscript.

Technique comments:

1. In the abstract, lines 19-20, the authors stated that "an innovative system", which cannot be supported by the results. An optimized system would be more understandable.

2. In the title, "A novel" is not realistic, most of the base models were referred from others. The title should be revised.

3. Line 41, "the reinforced concrete flat slabs"? Slabs are generally flat.

4. Line 73, what is the SFRC? It did not show before. Steel fiber-reinforced concrete?

5. The authors considered some factors of the SFRC slabs. However, the complicated material properties of SFRC were not well considered. There are some references (1. Experimental and Numerical Investigation of Fracture Behaviors of Steel Fiber-Reinforced Rubber Self-Compacting Concrete; 2. Fresh and mechanical performance and freeze-thaw durability of steel fiber-reinforced rubber self-compacting concrete (SRSCC)) may give some ideas related to the materials properties of SFRC, which could be included in the Introduction.

6. In Table 4, the "h" means？ I cannot find the definition of this factor.

7. The conclusions should be listed as several main subjects.

 

Author Response

Find the attached file. 

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

I accept the authors' replies. Several corrections and additions have been made to the manuscript. Overall, the quality of the manuscript has improved considerably. I recommend publication of the manuscript in this international journal.  

Author Response

Many thanks for your time to review our revision. 

Reviewer 3 Report

Thanks for the author's reply.

 

Some of the comments in the first review still needed to be addressed for the possible acceptance of this paper.

1. As mentioned in the first review, the introduction part should be revised, the material properties of the SFRC were not considered. Can this predicting model accurately predict different types of SFRC?

 

2. Line 22, the abstract should be corrected.

 

3. In Fig. 9, the title of the Y-axis should be revised, R^2, as well as Fig. 12.

 

4. In the conclusion, line 455， what engineering problem that you solved?

Author Response

Please see attached file. 

Author Response File: Author Response.pdf

Round 3

Reviewer 3 Report

The comments have been addressed.