Boosting Ensemble Learning for Freeway Crash Classification under Varying Traffic Conditions: A Hyperparameter Optimization Approach

Round 1

Reviewer 1 Report

A Boosting Ensemble Learning approach is proposed in the paper to enhance the accuracy of crash classification under varying traffic conditions. Real-world crash data from Flint, Michigan is used to train the models, which include five ensemble learning models (GradientBoosting, CatBoost, XGBoost, LightGBM, SGD). The models’ performance is further enhanced through hyperparameter optimization techniques. The evaluation results demonstrate that the proposed approach achieves a high accuracy rate of up to 99% in crash classification, particularly when Gradient Boost algorithms are implemented. Valuable insights into the potential of Boosting Ensemble Learning and its application in accurately classifying freeway crashes across different traffic conditions are provided by the paper. However, there are some issues in the article that need to be addressed:

(1)   It is suggested that the author address the limitations of statistical models in handling large and complex collision datasets.

(2)   It is suggested that the author explain the difficulties in validating linear and other robust statistical assumptions in real collision scenarios.

(3)   The author can elaborate on the necessity of utilizing non-parametric data mining techniques to uncover hidden patterns and interactions in massive datasets.

(4)   Has the author considered comparing the proposed method with existing methods in the field? If so, could you discuss the advantages and limitations of your method compared to these baselines?

(5)   Can the author provide more specific implementation details, such as preprocessing steps, feature selection, and the hyperparameters of the models used in the study?

(6)   Figure 1 is not referred to in the text; it is recommended that the author refer to figure 1.

(7)   The chart titles are inconsistent, such as “Figure.3.” and “Figure 4.”, please ask the author to unify them.

(8)   The paragraph formatting is inconsistent, with some paragraphs indented by 2 characters at the beginning, while others have no indentation.

(9)   Please correct the blurry text in some parts of Figure 3, as it is difficult to read.

(10) Please adjust the line spacing after Formula 1 in section 3.4.1 to match the line spacing in the rest of the document. Please ask the author to check the line spacing in other parts and ensure consistency.

(11) Please standardize the border format of Table 3 to match the other tables. Also, there are two bottom lines in Table 6, please correct it.

(12) There are still many formatting errors in the article. Please ask the author to double-check.

Moderate editing of English language required.

Author Response

 I wanted to thank you for taking the time to go through our manuscript and providing us with your insightful comments. Your feedback is incredibly valuable to us, and we've made some important updates based on your suggestions. Let me show you through the changes we've implemented:

Author Response File: Author Response.docx

Reviewer 2 Report

In the study, the authors aimed to enhance crash classification accuracy by incorporating crucial traffic factors like braking, weather conditions, and speed.Real-world crash data from Flint, Michigan is employed to assess the effectiveness of the proposed model. To achieve this goal, the authors embraced an innovative Boosting Ensemble Learning approach, harnessing five sophisticated ensemble learning models: GradientBoosting, CatBoost, XGBoost, LightGBM, and SGD. By leveraging hyperparameter optimization techniques, the authors further refine the performance of these models, elevating their overall predictive capabilities.The evaluation results showcase that the proposed approach, especially when implementing Gradient Boost algorithms, achieves an impressive accuracy rate of up to 99% in crash classification. This research unveils valuable insights into the potential of Boosting Ensemble Learning to accurately and efficiently classify freeway crashes across a range of traffic conditions. Furthermore, it sheds light on the nuanced variations in crash mechanisms observed when utilizing diverse ensemble learning models.By embracing the multifaceted nature of traffic conditions, this study paves the way for comprehensive strategies aimed at enhancing road safety and mitigating the impact of freeway crash. Generally, this is a good work. It can be accepted if the authors can consider the following issues: 1. What is the main motivation of the work? 2. Is the Hyperparameter Optimization Approach applicable for other kind of systems? 3. The quality of figure 2 should be improved. 4. More related works on the topics are welcome to enrich the literature review such as A Survey of Intelligent Driving Vehicle Trajectory Tracking Based on Vehicle Dynamics; Using an Inerter-Based Suspension to Reduce Carbody Flexible Vibration and Improve Riding-Comfort. 5. The language should be improved.

In the study, the authors aimed to enhance crash classification accuracy by incorporating crucial traffic factors like braking, weather conditions, and speed.Real-world crash data from Flint, Michigan is employed to assess the effectiveness of the proposed model. To achieve this goal, the authors embraced an innovative Boosting Ensemble Learning approach, harnessing five sophisticated ensemble learning models: GradientBoosting, CatBoost, XGBoost, LightGBM, and SGD. By leveraging hyperparameter optimization techniques, the authors further refine the performance of these models, elevating their overall predictive capabilities.The evaluation results showcase that the proposed approach, especially when implementing Gradient Boost algorithms, achieves an impressive accuracy rate of up to 99% in crash classification. This research unveils valuable insights into the potential of Boosting Ensemble Learning to accurately and efficiently classify freeway crashes across a range of traffic conditions. Furthermore, it sheds light on the nuanced variations in crash mechanisms observed when utilizing diverse ensemble learning models.By embracing the multifaceted nature of traffic conditions, this study paves the way for comprehensive strategies aimed at enhancing road safety and mitigating the impact of freeway crash. Generally, this is a good work. It can be accepted if the authors can consider the following issues: 1. What is the main motivation of the work? 2. Is the Hyperparameter Optimization Approach applicable for other kind of systems? 3. The quality of figure 2 should be improved. 4. More related works on the topics are welcome to enrich the literature review such as A Survey of Intelligent Driving Vehicle Trajectory Tracking Based on Vehicle Dynamics; Using an Inerter-Based Suspension to Reduce Carbody Flexible Vibration and Improve Riding-Comfort. 5. The language should be improved.

Author Response

 I wanted to thank you for taking the time to go through our manuscript and providing us with your insightful comments. Your feedback is incredibly valuable to us, and we've made some important updates based on your suggestions. Let me show you through the changes we've implemented:

Author Response File: Author Response.docx

Reviewer 3 Report

1. The abstract mentions using five ensemble learning models but doesn't clarify how they were selected or the rationale behind their choice. A brief explanation of why these models were chosen or their strengths in relation to the problem could add depth to the abstract.

2. The use of terms like "impressive accuracy rate of up to 99%" can be seen as overly optimistic and may raise expectations unreasonably. It's important to temper these claims with a discussion of potential limitations and the real-world applicability of such high accuracy.

3. The discussion and conclusion sections provide valuable insights but could benefit from more robust statistical analysis, a deeper exploration of practical implications, a thorough discussion of limitations, and careful use of language to convey the degree of certainty in the findings.

4. There are some grammatical and punctuation issues in the text, such as missing commas and a sentence that ends abruptly. Proper grammar and punctuation are essential for clarity and professionalism.

5. The final sentence in the abstract "mitigating the impact of freeway crashe," is incomplete and lacks clarity. It should be revised to provide a clear and complete thought.

Some grammatical errors are there, to be corrected.

Author Response

 I wanted to thank you for taking the time to go through our manuscript and providing us with your insightful comments. Your feedback is incredibly valuable to us, and we've made some important updates based on your suggestions. Let me show you through the changes we've implemented:

Author Response File: Author Response.docx

Reviewer 4 Report

The context and the scenario is appropriate following a case study method. Somewhere, the flow for reading is missing as the article goes back and forth with statistics and machine learning approaches for analysis. 

As we are living in the era of AI, we need to look at solutions to the problems or crashes in this case. That is more of prescriptive solution. What does the confusion matrix mean to avoid crash. This needs elaboration. 

Writing can be more crisp, to the point and succinct. Focus on the domain or the problem in hand needs improvement. 

In general, English language is fine. However, rewriting the article avoiding repeated sentences is recommended

Author Response

 I wanted to thank you for taking the time to go through our manuscript and providing us with your insightful comments. Your feedback is incredibly valuable to us, and we've made some important updates based on your suggestions. Let me show you through the changes we've implemented:

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Accept in present form.

Author Response

I wanted to thank you for taking the time to go through our manuscript and providing us with your insightful comments. Your feedback is incredibly valuable to us, and we've made some important updates based on your suggestions. Let me show you through the changes we've implemented:

Author Response File: Author Response.docx

Reviewer 4 Report

Modifications as per the suggestions are incorporated.

So, minor corrections in this case are related to crisp to-the-point writing for better readability.

English language flow is acceptable

Author Response

I wanted to thank you for taking the time to go through our manuscript and providing us with your insightful comments. Your feedback is incredibly valuable to us, and we've made some important updates based on your suggestions. Let me show you through the changes we've implemented:

Author Response File: Author Response.docx