A Hybrid Model Utilizing Principal Component Analysis and Artificial Neural Networks for Driving Drowsiness Detection
Marco Podda
Round 1
Reviewer 1 Report
Dear Authors,
I reviewed the study titled " A Hybrid Model of Data Mining and Artificial Neural Network for Driving Drowsiness Detection". The manuscript is generally well written. However, it has many technical, conceptual, and formatting deficiencies. In line with my comments, I think that it is not suitable for publication in Applied Science.
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The paper describes a method that combines PCA and ANNs to classify drowsiness based on physiological data of drivers. The evaluation is performed against other machine learning methods (SVM and KNN), as well as with/without PCA. The amount of work put forth in the paper looks impressive, but the experimental part lacks detail:
1) how where the hyper-parameters (e.g. learning rate of the BPNN) selected?
2) how come BPNN and CFNN use 20% data for validation, while SVM and KNN do not?
3) related to 2), the test dataset used to evaluate SVM and KNN is different from that used to evaluate BPNN and CFNN, which makes the results unreliable
4) still related to 2, hyper-parameters such as C for the SVM and k for KNN need to be tuned on a validation set. Given that you did not use a validation set, I assume they were not tuned. In contrast, I don't know if the same applies to BPNN and CFNN.
In light of this, I suggest the authors answer my questions to improve the quality of the manuscript.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
I attached my comments.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
I find the response of the authors satisfactory, and I recommend acceptance.
Author Response
Dear reviewer:
Thank you for your recommendation.
Round 3
Reviewer 1 Report
It has been sufficiently improved to warrant publication in Applied Sciences.
