Prediction of Ship CO₂ Emissions and Fuel Consumption Using Voting-BRL Model

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript proposed a Voting-BRL model to predict ship CO2 emissions and fuel consumption. This study is interesting and significant for the effective ship emission management and operational efficiency. However, there are some problems that would seem lead to a major revision. 

1.       The novelty and advantage of the model proposed in this study is doubtful. The description of model and its experiment is quite simple. The novelty and advantage of the model is not well proved and analyzed.

2.       The related works is simple and does not provide convincing evidence of the novelty and advantage of the model proposed. Furthermore, it should focus on the prediction of ship CO2 emissions and fuel consumption and the methods applied.

3.       The challenges include nonlinear characteristics of fuel consumption and emission data, the variability of the marine environment, etc. This study claims “To address these challenges, this study proposes a Voting Regressor model”. However, the model proposed in this study also has not address these issues.

4.       This study employs ANOVA to select features. But this manuscript does not demonstrate how it select/discard the features from original datasets. The process and results of ANOVA keeps unclear.

5.       There is not any information about the datasets for experiments, especially the volume, parameters of data. Detailed information of dataset would guarantee the trustworthiness of the model and its experiments.

6.       Furthermore, do the predictions of ship CO2 emissions and fuel consumption share the same model? Same dataset? Same experiment? Same discussion? The ship CO2 emissions and fuel consumption are different. But this manuscript did not deal with them respectively.

7.       “marine environment—such as weather conditions, route choices, and ship loading—further increases the complexity of the data and the difficulty of prediction“. But route choices, and ship loading are not marine environment.

8.       “As computing power increases, the application of Bayesian methods is expected to grow…. ”, It’s so confusing. Bayesian methods consume huge power to run?

“Lasso’s application is expected to increase as the volume and complexity of shipping data continue to grow”. Those statements are not logical. The reason for the increase of application of Lasso contributes to the advantages of Lasso, rather to the shipping data.

9.       It’s advisable that figures appear after/close to their first citation, such as Figure. 1.

10.   The captions of figures are quite small and unreadable, Figure 1~7.

11.   VE-BRL in figure 1 = Voting-BRL?

Comments on the Quality of English Language

 The English could be improved to more clearly express the research.

Author Response

This manuscript proposed a Voting-BRL model to predict ship CO2 emissions and fuel consumption. This study is interesting and significant for the effective ship emission management and operational efficiency. However, there are some problems that would seem lead to a major revision.

1.The novelty and advantage of the model proposed in this study is doubtful. The description of the model and its experiment is quite simple. The novelty and advantage of the model is not well proved and analyzed.

A：Thank you for your valuable feedback. We appreciate the opportunity to clarify the novelty and advantages of the model proposed in this study. While deep learning models, such as those utilizing layer-by-layer feature extraction, have demonstrated promising results and represent the innovation in many recent works, we would like to emphasize that traditional regression models still maintain high accuracy when modeling features directly. This area of research continues to offer significant advantages.

In fact, our comparative experiment (Table 3) demonstrates the performance of the MLP Regressor as a representative of deep learning-based regression. The proposed Voting-BRL model outperforms MLP Regressor significantly, improving the R² score from -0.2 to 0.9981. This clearly highlights the effectiveness and advantages of our approach in comparison to deep learning models for this specific task.

We believe that our method offers a valuable contribution to the field by combining the strengths of both traditional regression and modern ensemble learning techniques, providing robust and accurate predictions.

The revised are as follows:

In the context of increasing global warming and heightened environmental awareness, the issues of carbon dioxide (CO2) emissions in the shipping industry have garnered extensive attention. As one of the major sources of global greenhouse gas emissions, the maritime transportation sector faces growing challenges in reducing its carbon footprint amidst the continued expansion of global trade. According to data from the International Maritime Organization (IMO), the shipping industry accounts for approximately 2.5\% of global greenhouse gas emissions, and this share is projected to rise as international trade volumes increase \cite{song2023dynamic, ha2023framework, pelic2023impact}. Improving the operational efficiency of ships and minimizing their carbon emissions have therefore become critical goals for sustainable development in the shipping sector.

Accurately predicting ship CO2 emissions is essential to achieving these goals. However, this task is inherently complex due to the nonlinear nature of emission data. Traditional linear regression models struggle to adequately capture the intricate relationships among the various influencing factors, leading to limited predictive accuracy\cite{damartzis2022solvents, einbu2022energy}. Moreover, the inherent high-dimensionality and heterogeneity of the data exacerbate these challenges, as they introduce noise and make it difficult for conventional models to generalize effectively. Current machine learning approaches, such as Random Forest, Support Vector Machines (SVM)\cite{zhang2023framework, gordon2023support}, and Neural Networks\cite{khoshraftar2023modeling, sedighi2025comparative}, have demonstrated improved predictive performance compared to traditional methods\cite{zhang2023use, de2023prediction, khoshraftar2023modeling}. However, these methods also exhibit significant limitations. They often require large quantities of high-quality data, are computationally intensive, and lack transparency due to their "black-box" nature, making it challenging to interpret how specific features contribute to the predictions\cite{nagao2024efficient, lv2023modelling, nguyen2023extensive}. Furthermore, in scenarios with limited or noisy data, these models are prone to overfitting, which further diminishes their generalization capabilities.

Existing linear regression models also face significant challenges in this domain. While these models offer greater interpretability and lower computational costs compared to machine learning methods, they often fail to handle nonlinear and high-dimensional feature data effectively. Despite the application of preprocessing techniques such as feature scaling, traditional regression approaches remain highly susceptible to interference from irrelevant features, which degrades their prediction accuracy. Additionally, these models lack mechanisms for effective feature filtering, such as those found in deep learning techniques, further limiting their predictive capabilities in complex scenarios. To address these challenges, this study proposes a Voting Regressor Model (Voting-BRL) to enhance the accuracy and robustness of ship CO2 emissions predictions. The proposed method first employs Analysis of Variance (ANOVA) to select features highly correlated with the target variables, thereby reducing the dimensionality of the data, decreasing model complexity, and mitigating the impact of irrelevant features. Next, a combination of Bayesian Ridge Regression and Lasso Regression is utilized. Bayesian Ridge Regression is effective in handling uncertainty and multicollinearity, while Lasso Regression excels at automatic feature selection, further improving predictive performance and robustness. Finally, a voting mechanism is used to integrate the predictions of the two models, enhancing the generalization ability of the model and reducing the risk of overfitting in scenarios with limited or noisy data.

 

2. The related works is simple and does not provide convincing evidence of the novelty and advantage of the model proposed. Furthermore, it should focus on the prediction of ship CO2 emissions and fuel consumption and the methods applied.

        A：Thank you for your insightful comments. In response to your feedback, we have revised the related works section to provide a more detailed and comprehensive discussion. We have included additional references to better illustrate the state-of-the-art approaches in the prediction of ship CO₂ emissions and fuel consumption, as well as the methods applied in this area.

In particular, we have expanded on the comparison between traditional and modern approaches, highlighting how the proposed model addresses existing challenges and improves upon previous work. These modifications aim to better demonstrate the novelty and advantages of our model.

 

 

3. The challenges include nonlinear characteristics of fuel consumption and emission data, the variability of the marine environment, etc. This study claims “To address these challenges, this study proposes a Voting Regressor model”. However, the model proposed in this study also has not address these issues.

A：Thank you for your valuable feedback. We appreciate the opportunity to address the concerns raised regarding the challenges of nonlinear characteristics of fuel consumption and emission data, as well as the variability of the marine environment.

In response to your comment, we have revised our description to clarify that while the Voting Regressor model proposed in this study does not directly tackle all aspects of these challenges, it significantly improves the accuracy and robustness of predictions under varying conditions. Specifically, the model enhances the prediction of fuel consumption and CO₂ emissions by leveraging the strengths of multiple regression models, providing a more stable and reliable performance despite the nonlinearities and environmental variability.

 

 

4. This study employs ANOVA to select features. But this manuscript does not demonstrate how it select/discard the features from original datasets. The process and results of ANOVA keeps unclear.

A：Thank you for your feedback. In response to your comment, we have added a more detailed explanation of the feature selection process using ANOVA. Specifically, we have clarified how features are selected or discarded from the original dataset based on their statistical significance. This explanation is now included in the manuscript to ensure the process and results of ANOVA are clear. We believe this additional clarification will help address the concerns regarding the transparency of the feature selection process.

The revised are as follows:

The overall structure of our model is illustrated in Figure 1. In this framework, we first apply an ANOVA (Analysis of Variance) technique for feature selection, then perform regression using a hybrid method that combines Bayesian Ridge Regression and Lasso Regression, ultimately outputting final predictions through an ensemble learning strategy known as Voting Regressor. In Figure11, first, categorical variables in the dataset are converted to numerical representations using one-hot encoding, resulting in 3,610 features. Relevant features are then selected through ANOVA analysis, which evaluates the correlation between each feature and the target variable. This process reduces the feature set to 100 highly relevant features, effectively minimizing noise and dimensionality while retaining the critical information for regression. Subsequently, regression is performed using the Voting-BRL method to output predictions.

5.There is not any information about the datasets for experiments, especially the volume, parameters of data. Detailed information of dataset would guarantee the trustworthiness of the model and its experiments.

A：Thank you for your valuable feedback. In response to your comment, we have revised the manuscript to include detailed information about the datasets used for the experiments. This includes the volume, parameters, and other relevant characteristics of the data. We believe that providing this information will enhance the transparency and trustworthiness of the model and its experimental results.

The revised are as follows:

The data originates from the THETIS-MRV platform managed by the European Maritime Safety Agency (EMSA), which focuses on the monitoring, reporting, and verification (MRV) system of ship emissions. This platform provides publicly accessible ship emission data, enabling users to view and analyze CO2 emissions from ships operating within EU waters. The dataset contains records from 2020 to 2023, with the following number of data points per year after accounting for data exclusions:

2020: 12,117 records,

2021: 12,485 records,

2022: 13,468 records,

2023: 12,439 records.

These features include fuel consumption (measured in tons), navigation distance (in nautical miles), and CO2 emissions (in tons), providing a comprehensive foundation for developing predictive models and ensuring transparency in environmental compliance within the shipping industry. Table \ref{tab:data_info} presents a summary of the dataset's key characteristics.

 

6. Furthermore, do the predictions of ship CO2 emissions and fuel consumption share the same model? Same dataset? Same experiment? Same discussion? The ship CO2 emissions and fuel consumption are different. But this manuscript did not deal with them respectively.

A：Thank you for pointing out this issue. We apologize for the oversight in the manuscript. The reference to fuel consumption was a typographical error. Our study focuses solely on the prediction of ship CO\textsubscript{2} emissions, and we do not address fuel consumption. We have carefully reviewed and revised the manuscript to ensure that all instances of fuel consumption have been corrected or removed to accurately reflect the scope of our research.

 

 

7. “marine environment—such as weather conditions, route choices, and ship loading—further increases the complexity of the data and the difficulty of prediction“. But route choices, and ship loading are not marine environment.

A：Thank you for your insightful comment. We acknowledge that "route choices" and "ship loading" are not part of the marine environment. We have revised this section of the manuscript to ensure a more accurate and precise description.

 

 

8.“As computing power increases, the application of Bayesian methods is expected to grow…. ”, It’s so confusing. Bayesian methods consume huge power to run? “Lasso’s application is expected to increase as the volume and complexity of shipping data continue to grow”. Those statements are not logical. The reason for the increase of application of Lasso contributes to the advantages of Lasso, rather to the shipping data.

A：Thank you for your thoughtful feedback. We understand that the original statements could be misleading and lacked clarity. We have revised this section of the manuscript to provide a more accurate explanation.

 

9.It’s advisable that figures appear after/close to their first citation, such as Figure. 1.

A：Thank you for pointing out the issue regarding the placement of figures. We have revised the layout of the manuscript to ensure that all figures, including Figure 1, now appear immediately after or close to their first citation. This adjustment enhances the readability and logical flow of the manuscript.

 

 

10.The captions of figures are quite small and unreadable, Figure 1~7.

A：Thank you for your valuable feedback regarding the figure captions. We have carefully revised and reformatted all figures (Figure 1–7) to ensure that the captions are now more legible. The font size has been adjusted, and additional spacing has been incorporated to enhance readability.

 

 

11. VE-BRL in figure 1 = Voting-BRL?

A：Thank you for pointing out the inconsistency in the terminology. We have updated Figure 1 and ensured that "VE-BRL" has been replaced with "Voting-BRL" for consistency throughout the manuscript.

 

 

Reviewer 2 Report

Comments and Suggestions for Authors

In the introduction, a brief description of how it feeds the model.

I would relocate the contributions of the article to the conclusions.

In section 2.1, there was no talk of CO2 capture systems; I suggest some references:

https://doi.org/10.1016/j.ijggc.2021.103526

https://doi.org/10.3390/app12126100

Concatenate section 2.1 with the problem to be solved.

Locate Figure 1 after it is named in the manuscript, and the explanation of the figure should be in the document's body.

Change the title of Figure 1 to a shorter one.

It would be interesting if Figure 1 showed in detail the variables taken by the model.

There is a table with no title between lines 247 and 248. Likewise, the information contained in this table (algorithm 1) would be very helpful to the reader if it were presented as a flow chart.

Tables 1 and 2 should appear after being mentioned in the document.

The first time RMSE is mentioned in the document's body, its meaning should be written in full, and its acronym must be in parentheses, as in the abstract.

Table 3, time in which units?

Separate Figure 2 for better understanding. Change the upper right graph in the same figure to another type of graph or adjust the scale to highlight the differences between the models.

Reduce the title in Figure 2 and introduce its explanation in the document's body.

Figure 3. What are the blue dots? What variable is CO2 or fuel consumption?

What is the difference between Tables 3 and 4 for the year 2023?

If the results are the same, why do these models present the same values?

Section 5.1 What parts of the model were modified or eliminated for the experiments?

The fuel consumption model is not visible anywhere.

Formatting corrections

Write in the third person.

CO2 as a subscript throughout the document

The authors' contribution, funding, etc., should be completed at the end of the manuscript.

Improve the figures in section 5.2 to make them more visible.

 

In general, the article shows the development of a model to predict CO2 emissions and fuel consumption more accurately. Although the results obtained with the proposed model appear more accurate than those of other models, they do not show which variables most affect the model, and there is no deep discussion about this. On the other hand, it is unclear how the proposed model would mitigate CO2 emissions and fuel consumption in the transport sector.

It would be interesting to show predictions of how CO2 and fuel consumption would vary with different models, for example, until 2030 or 2050. This analysis would show how CO2 would be mitigated and fuel consumption reduced.

Author Response

 

1.In the introduction, a brief description of how it feeds the model. I would relocate the contributions of the article to the conclusions.

A: Thank you for your insightful suggestion. We have revised the introduction to include a brief description of how the data feeds into the model, providing a clearer context for the proposed methodology. Additionally, we have relocated the detailed contributions of the article to the conclusion section, ensuring a more logical flow and alignment with standard practices.

The revised are as follows:

In the context of increasing global warming and heightened environmental awareness, the issues of carbon dioxide (CO2) emissions in the shipping industry have garnered extensive attention. As one of the major sources of global greenhouse gas emissions, the maritime transportation sector faces growing challenges in reducing its carbon footprint amidst the continued expansion of global trade. According to data from the International Maritime Organization (IMO), the shipping industry accounts for approximately 2.5\% of global greenhouse gas emissions, and this share is projected to rise as international trade volumes increase \cite{song2023dynamic, ha2023framework, pelic2023impact}. Improving the operational efficiency of ships and minimizing their carbon emissions have therefore become critical goals for sustainable development in the shipping sector.

Accurately predicting ship CO2 emissions is essential to achieving these goals. However, this task is inherently complex due to the nonlinear nature of emission data. Traditional linear regression models struggle to adequately capture the intricate relationships among the various influencing factors, leading to limited predictive accuracy\cite{damartzis2022solvents, einbu2022energy}. Moreover, the inherent high-dimensionality and heterogeneity of the data exacerbate these challenges, as they introduce noise and make it difficult for conventional models to generalize effectively. Current machine learning approaches, such as Random Forest, Support Vector Machines (SVM)\cite{zhang2023framework, gordon2023support}, and Neural Networks\cite{khoshraftar2023modeling, sedighi2025comparative}, have demonstrated improved predictive performance compared to traditional methods\cite{zhang2023use, de2023prediction, khoshraftar2023modeling}. However, these methods also exhibit significant limitations. They often require large quantities of high-quality data, are computationally intensive, and lack transparency due to their "black-box" nature, making it challenging to interpret how specific features contribute to the predictions\cite{nagao2024efficient, lv2023modelling, nguyen2023extensive}. Furthermore, in scenarios with limited or noisy data, these models are prone to overfitting, which further diminishes their generalization capabilities.

Existing linear regression models also face significant challenges in this domain. While these models offer greater interpretability and lower computational costs compared to machine learning methods, they often fail to handle nonlinear and high-dimensional feature data effectively. Despite the application of preprocessing techniques such as feature scaling, traditional regression approaches remain highly susceptible to interference from irrelevant features, which degrades their prediction accuracy. Additionally, these models lack mechanisms for effective feature filtering, such as those found in deep learning techniques, further limiting their predictive capabilities in complex scenarios. To address these challenges, this study proposes a Voting Regressor Model (Voting-BRL) to enhance the accuracy and robustness of ship CO2 emissions predictions. The proposed method first employs Analysis of Variance (ANOVA) to select features highly correlated with the target variables, thereby reducing the dimensionality of the data, decreasing model complexity, and mitigating the impact of irrelevant features. Next, a combination of Bayesian Ridge Regression and Lasso Regression is utilized. Bayesian Ridge Regression is effective in handling uncertainty and multicollinearity, while Lasso Regression excels at automatic feature selection, further improving predictive performance and robustness. Finally, a voting mechanism is used to integrate the predictions of the two models, enhancing the generalization ability of the model and reducing the risk of overfitting in scenarios with limited or noisy data.

 

In this study, we introduced the Voting-BRL model, an innovative ensemble learning approach that integrates Bayesian Ridge Regression and Lasso Regression, to predict ship carbon dioxide (CO\textsubscript{2}) emissions and fuel consumption with high accuracy and robustness. By leveraging Analysis of Variance (ANOVA) for feature selection, the model effectively reduced dimensionality and minimized noise interference, enhancing its predictive performance. Experimental results demonstrated that Voting-BRL achieved an outstanding $R^2$ of 0.9981 and a Root Mean Square Error (RMSE) of 8.53, markedly outperforming traditional machine learning models such as XGBRegressor, which attained an $R^2$ of 0.97 and an RMSE of 45.03. Ablation studies confirmed that the ensemble strategy harnesses the complementary strengths of Bayesian Ridge and Lasso Regression, resulting in superior generalization capabilities and prediction stability.

The exceptional performance of the Voting-BRL model underscores its potential as a reliable tool for emission management and operational optimization within the maritime industry. Accurate predictions of CO\textsubscript{2} emissions is crucial for developing strategies to enhance environmental sustainability and comply with increasingly stringent regulatory standards. By providing precise forecasts, the Voting-BRL model can assist stakeholders in making informed decisions that contribute to reducing the carbon footprint of shipping operations. The main contributions of this study include:

Proposed the Voting-BRL (Voting-Bayesian Ridge and Lasso) method: This method combines Bayesian Ridge Regression and Lasso Regression through a voting mechanism to achieve more precise carbon dioxide emission predictions.

Conducted detailed ablation experiments: These experiments analyze the impact of different modules on the performance of the Voting-BRL model across multiple datasets, validating the effectiveness of each component of the model.

Validated the method using real-world data: Utilizing four years of actual data from the THETIS-MRV platform managed by the European Maritime Safety Agency (EMSA), the experimental results demonstrate that the Voting-BRL model achieves or exceeds an $R^2$ of 0.99 in prediction performance, significantly outperforming traditional methods and showcasing its efficiency and reliability in practical applications.

Future work may focus on expanding the model to incorporate additional environmental and operational factors, thereby further enhancing its predictive accuracy and applicability. Additionally, integrating real-time data streams could enable dynamic emission monitoring and adaptive decision-making in response to changing maritime conditions. Exploring the application of the Voting-BRL framework to other sectors within the transportation industry may also yield valuable insights and broaden its impact on global efforts to mitigate greenhouse gas emissions.

 

 

2.In section 2.1, there was no talk of CO2 capture systems; I suggest some references: https://doi.org/10.1016/j.ijggc.2021.103526 ;https://doi.org/10.3390/app12126100

A: Thank you for your valuable suggestion. We have incorporated references related to CO2 capture systems in Section 2.1 to enhance the discussion. The following references have been added to provide a more comprehensive overview

 

 

3.Concatenate section 2.1 with the problem to be solved.

A：Thank you for the suggestion. We have revised the manuscript to concatenate Section 2.1 with the problem statement, ensuring a more cohesive and logical flow.

 

 

4.Locate Figure 1 after it is named in the manuscript, and the explanation of the figure should be in the document's body.

A：Thank you for your feedback. We have repositioned Figure 1 to appear immediately after its first mention in the manuscript and integrated a detailed explanation of the figure into the main text for better clarity and alignment.

 

 

5.Change the title of Figure 1 to a shorter one.

A：Thank you for your suggestion. We have revised the title of Figure 1 to make it shorter and more concise, ensuring it better aligns with the manuscript’s content.

 

 

6.It would be interesting if Figure 1 showed in detail the variables taken by the model.

A：Thank you for the suggestion. We have updated Figure 1 to include a more detailed representation of the variables taken by the model, enhancing its clarity and providing a better understanding of the model’s components.

 

 

7.There is a table with no title between lines 247 and 248. Likewise, the information contained in this table (algorithm 1) would be very helpful to the reader if it were presented as a flow chart.

A：hank you for your feedback. We have added a detailed explanation of the table in the text. Additionally, to improve clarity, we have presented the information from Algorithm 1 as a flowchart, making it easier for the reader to follow the process.

 

 

8.Tables 1 and 2 should appear after being mentioned in the document.

A：Thank you for your suggestion. We have rearranged the manuscript so that Tables 1 and 2 now appear immediately after they are first mentioned in the text, ensuring better alignment with the document’s flow.

 

9.The first time RMSE is mentioned in the document's body, its meaning should be written in full, and its acronym must be in parentheses, as in the abstract.

A：Thank you for your suggestion. We have updated the manuscript to write out the full form of RMSE (Root Mean Squared Error) the first time it is mentioned in the document, with the acronym in parentheses, consistent with the style used in the abstract.

 

 

10.Table 3, time in which units?

A：Thank you for pointing that out. We have updated Table 3 to specify that the time is measured in seconds (s), ensuring clarity for the reader.

 

 

11.Separate Figure 2 for better understanding. Change the upper right graph in the same figure to another type of graph or adjust the scale to highlight the differences between the models.

A：Thank you for your suggestion. We have separated Figure 2 into distinct figures for better clarity.

 

 

12.Reduce the title in Figure 2 and introduce its explanation in the document's body.

A：Thank you for your feedback. We have reduced the title in Figure 2 to make it more concise and have incorporated its detailed explanation into the body of the document, as suggested.

 

 

13.Figure 3. What are the blue dots? What variable is CO2 or fuel consumption?

A：Thank you for your comment. The blue dots in Figure 3 represent CO₂ emissions. We have clarified this in the manuscript, where we explicitly mention that the blue dots correspond to CO₂ emissions.

 

 

14.What is the difference between Tables 3 and 4 for the year 2023? If the results are the same, why do these models present the same values?

A：Thank you for your question. The difference between Tables 3 and 4 for the year 2023 lies in the type of experiments conducted. Table 4 presents the results of an ablation study, where we tested the model's performance by systematically removing certain components. Despite these variations, the results from Table 3 and Table 4 remain consistent, confirming the robustness of our model across different setups.

 

 

16.Section 5.1 What parts of the model were modified or eliminated for the experiments?

A：The reason for the identical results in both tables is due to the ablation experiment presented in Table 4. In this experiment, we systematically removed certain components from the model to test their impact on performance. Despite removing parts of the model, the results remain consistent, which highlights the robustness of the proposed approach. This has been clearly indicated in the manuscript.

 

 

17.The fuel consumption model is not visible anywhere.

A：Thank you for your comment. The mention of fuel consumption was a typographical error, and we have removed it from the manuscript. The focus of our study is on CO2 emissions prediction, and this has been clarified throughout the paper.

 

 

18.Formatting corrections. Write in the third person. CO2 as a subscript throughout the document

A: Thank you for your feedback. We have made the necessary formatting corrections, including rewriting the manuscript in the third person and ensuring that "CO2_22​" is consistently written as a subscript throughout the document.

19.The authors' contribution, funding, etc., should be completed at the end of the manuscript.

A: Thank you for the suggestion. We have moved the authors' contribution, funding, and other relevant information to the end of the manuscript as requested.

 

 

20.Improve the figures in section 5.2 to make them more visible.

A: We have improved the figures in Section 5.2 to enhance their visibility, ensuring better clarity and readability.

 

 

21.In general, the article shows the development of a model to predict CO2 emissions and fuel consumption more accurately. Although the results obtained with the proposed model appear more accurate than those of other models, they do not show which variables most affect the model, and there is no deep discussion about this. On the other hand, it is unclear how the proposed model would mitigate CO2 emissions and fuel consumption in the transport sector.

A: Thank you for your comments. We have revised the manuscript to provide a deeper discussion on the key variables that significantly affect the performance of the proposed model. Additionally, we have clarified how the model contributes to mitigating CO2 emissions in the transport sector, specifically emphasizing its potential impact on decision-making and optimization in shipping operations. We hope these revisions address your concerns and provide a clearer understanding of the model's practical implications.

 

 

22.It would be interesting to show predictions of how CO2 and fuel consumption would vary with different models, for example, until 2030 or 2050. This analysis would show how CO2 would be mitigated and fuel consumption reduced.

A: Thank you for your suggestion. Unfortunately, the current dataset does not include future data for the years 2030-2050, which limits our ability to make long-term predictions regarding CO2 emissions. As a result, we are unable to perform the requested analysis for future years. However, we have focused on providing accurate predictions based on the available historical data, and we can explore future trends in subsequent studies as more data becomes available.

 

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

None