The Impact of Partial Balance of Imbalanced Dataset on Classification Performance
Round 1
Reviewer 1 Report
First of all, the paper “The Impact of Partial Balance of Imbalanced Dataset on Classification Performance” aims and scope match those of Electronics. However based on my opinion it needs minor improvements to be considered for publication in Electronics. I would suggest the following changes that in my opinion would improve the paper, in special for the reader.
- The abstract is loosely written. It is not as informative as expected. A standard abstract must present, without leaving any doubt, the objective of the paper precisely; source of data (which is not present in your abstract) and analytical approach used; key findings and any policy implication and recommendations.
- I suggest the authors to improve the introduction section. Authors should better highlight the objective of their work and to what extent it contributes to close a gap in the existing literature and/or practice. What is the innovative value of the contribution proposed by the authors?
- all variables used in equations should be introduced in the main text. Equations (7) – (9) – variables are not explained. The similar situation is with other variables.
- Validation section is not well prepared. How we can judge about these results? Comparisons with existing algorithms from the literature is missing.
- The conclusion section seems to rush to the end. The authors will have to demonstrate the impact and insights of the research. The authors need to clearly provide several solid future research directions. Clearly state your unique research contributions in the conclusion section. Add limitations. No bullets should be used in your conclusion section.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
# Authors does not provided the main contributions of this paper.
# Literature work is poorly reported in this paper
# Conclusion is very lengthy
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Title of Manuscript: The Impact of Partial Balance of Imbalanced Dataset on Classification Performance (electronics-1657966)
The manuscript presents a comparative study on imbalance dataset classification using two datasets and some parameter setting techniques. Author proposed novel parameters with detail explanation of their impact on classification accuracy in some cases of imbalance datasets. The study is interesting and widely applicable. However, the manuscript lacks novelty and is bias in several aspects.
Major comments
- The author used only one classifier (i.e. Random forest) in their experiment and tested differently cased on proposed parameters. It is too early to make a conclusion based on a single classifier for parameter selection for an imbalanced dataset. I strongly recommend the author include some machine learning and deep learning classifier to make the proposed procedure more general and meaningful.
- Two datasets (Moore and CICIDS2017) are used in this experiment and treated in various scenarios (as described in Figure 1). As far I know, each dataset is subjective. I am curious that how the author treated them in various cases of imbalance and balanced datasets. Therefore, it is important to include more datasets in the experiment.
- The manuscript provides less detail of the classifier and experiment setting. The author should provide some detail of their parameter setting in the classifier and if possible release the code detail for better understanding to the reader.
Minor comments
- The content is repeated on page 4 lines 151 to 160.
- Please make uniformity on abbreviation in line 246on page 4.
- Too many figures and it is hard to make a comparison on proper parameter selection. I recommend including some tricky figures for better understanding.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
# After revision, study provided is not found with novelty of work
# Dataset consider for experimentation was older, not enough to support the same.
# References written and cites has also not supporting the observations reported in this work.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
The representation of the results and conclusion should be improved.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf