In-Vehicle Data for Predicting Road Conditions and Driving Style Using Machine Learning
Round 1
Reviewer 1 Report
The authors used three supervised machine learning algorithms to predict driving behavior and road conditions using seven tagged CAN data: random forest, decision tree, and support vector machine. These methods were used to categorize the state of the road surface as either smooth, even, or devoid of holes. Here are my observations:
1- What is the study's main motivation? Why did the authors employ such basic techniques? why do they select these techniques? Are these techniques more effective than CNN or RNN techniques?
2-The introduction section lacks the phrases why, background, what, and gap. The writers should rewrite this section, it is advised.
3-The introduction section lacks important references. Authors are encouraged to reference and use these works.
https://www.sciencedirect.com/science/article/abs/pii/S2210670722004061
https://www.sciencedirect.com/science/article/pii/S0010482522002530
https://link.springer.com/article/10.1007/s00521-022-07424-w
4-Do not leave any sections empty. Complete section 7 with concise sentences describing the subsections and what happened afterward.
5-Formulas borrowed from other works must be properly referenced.
6-The authors ought to take Figures out of the conclusion section.
7-The conclusion section should discuss the implications of the study and the limitations of the method.
8-Include further details about the Figures.
9-The discussion section is not wide enough. Add several figures and tables to clearly explain your findings.
Author Response
Thank for reviewing the manuscript, the comments were very helpful to improve the work. Please find the attached PDF for the responses to your reviews
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors present the results of the implementation of 3 supervised learning algorithms (DT, RF, SVM) applied to the taken-over data set. Labeled data make it possible to classify road surface conditions (3 categories - full of holes, smooth, even), road traffic conditions (3 categories - low, normal, and high traffic), and the driving style (2 categories - aggressive and normal).
140: There is no justification as to why just these algorithms have been chosen and not any others. Criteria of selection?
145: The used description of Figure 1 (an explanatory part) would be better used in the accompanying text when referring to the figure. Thus Figure 1 could have a shorter and more apt name.
A comparison of the achieved results with the results of other authors is missing. Are there any applicable comparative studies?
From a formal point of view, the manuscript needs a lot of improvements. E.g., the quality of the manuscript suffers from incorrect referencing, e.g. in the text, there are no references to the literature [18] and [29], and there are two references [19] contained in the reference list. Some parts declared in the template have not been filled in yet, still containing general instructions to the authors (300-307, 309-318, 320-324), visibility of the last author’s Orcid number in the list of authors, etc., etc.
Extensive editing of the English language and style are required. To demonstrate this opinion, please, go through the language corrections proposed for the first 100 lines of the manuscript:
1: Vehicles transfer data on their network to allow efficient implementation of vehicle systems…
8: …road surface conditions…
16: Vehicles architecture has developed…
18: The number of ECUs in vehicles was increased…
21-22: For example, the Advanced Cruise Control system…
23: …object in the path…
28-29: …with each other.
30-31: Flexray is another vehicle network protocol that provides more bandwidth than CAN [3], allowing more connected ECUs.
37: …and much more important information…
37-38: This data is called in-vehicle data…
41: The idea of using In-vehicle data with machine learning (ML) has increased…
45: …using a mono camera…
46: Fusion data from many sensors like cameras and radars…
50: …for object detection…
51-52: This research focuses on implementing some safety features without the need for dedicated sensors to feed into the ML system.
52-53: The dataset used in this work was gathered…
58: Section 5 provides details…
59: …and the hyperparameters…
60: Section 7 presents the detection results.
68: The in-vehicle data recorder…
76: …distance plot.
78: …techniques, and the…
85: …machine (SVM) and…
88: A random forest classifier is used…
89: …a machine learning-based model…
92: …models were used… (OR …models are used…)
93: …to detect a distracted driver… (OR … to detect distracted drivers…)
100: …in-vehicle network security …
Author Response
Thanks for reviwing our manuscript. Your comments and feedback helped us improving the work.
Please find the responses for your reviews in the attached PDF
Regards,
Ghaith Al-refai
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
It can be accepted.
Author Response
Thank you very much for your efforts and reviews, They were very helpful to improve our work.
Regards
The Authors
Reviewer 2 Report
Substitute (Line 126): "3. system overview" with "3. System overview"
Author Response
Thank you very much for your efforts and reviews, They were very helpful to improve our work.
The section name is corrected as required "System Overview"
Regards
The Authors
