Enhancing PM2.5 prediction by mitigating annual data drift using wrapped loss and neural networks

Md Khalid Hossen; Yan-Tsung Peng; Meng Chang Chen

doi:10.1371/journal.pone.0314327

Peer Review History

Original SubmissionFebruary 12, 2024
10 Mar 2024 Decision Letter - Worradorn Phairuang, Editor PONE-D-24-05738On the Data Shifting: Predicting PM2.5 levels by using Enhanced Deep neural networkPLOS ONE Dear Dr. Hossen, Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. Please submit your revised manuscript by Apr 24 2024 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. Please include the following items when submitting your revised manuscript: A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols. We look forward to receiving your revised manuscript. Kind regards, Worradorn Phairuang, Ph.D. Academic Editor PLOS ONE Journal Requirements: When submitting your revision, we need you to address these additional requirements. 1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf 2. Please note that PLOS ONE has specific guidelines on code sharing for submissions in which author-generated code underpins the findings in the manuscript. In these cases, all author-generated code must be made available without restrictions upon publication of the work. Please review our guidelines at https://journals.plos.org/plosone/s/materials-and-software-sharing#loc-sharing-code and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse. 3. Thank you for stating the following financial disclosure: "This document is the result of the research project funded by the National Science and 363 Technology Council of Taiwan, under the grant NSC 109-2119-M-001-010-A." Please state what role the funders took in the study. If the funders had no role, please state: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript." If this statement is not correct you must amend it as needed. Please include this amended Role of Funder statement in your cover letter; we will change the online submission form on your behalf. 4. Please note that funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form. Please remove any funding-related text from the manuscript. 5. When completing the data availability statement of the submission form, you indicated that you will make your data available on acceptance. We strongly recommend all authors decide on a data sharing plan before acceptance, as the process can be lengthy and hold up publication timelines. Please note that, though access restrictions are acceptable now, your entire data will need to be made freely accessible if your manuscript is accepted for publication. This policy applies to all data except where public deposition would breach compliance with the protocol approved by your research ethics board. If you are unable to adhere to our open data policy, please kindly revise your statement to explain your reasoning and we will seek the editor's input on an exemption. Please be assured that, once you have provided your new statement, the assessment of your exemption will not hold up the peer review process. 6. Please amend either the abstract on the online submission form (via Edit Submission) or the abstract in the manuscript so that they are identical. 7. Please ensure that you refer to Figure 12 in your text as, if accepted, production will need this reference to link the reader to the figure. 8. We note you have included a table to which you do not refer in the text of your manuscript. Please ensure that you refer to Tables 4 and 8 in your text; if accepted, production will need this reference to link the reader to the Table. Additional Editor Comments: Major revision. [Note: HTML markup is below. Please do not edit.] Reviewers' comments: Reviewer's Responses to Questions Comments to the Author 1. Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented. Reviewer #1: Partly Reviewer #2: Partly ******** 2. Has the statistical analysis been performed appropriately and rigorously? Reviewer #1: Yes Reviewer #2: Yes ****** 3. Have the authors made all data underlying the findings in their manuscript fully available? The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified. Reviewer #1: Yes Reviewer #2: Yes ****** 4. Is the manuscript presented in an intelligible fashion and written in standard English? PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here. Reviewer #1: Yes Reviewer #2: Yes ****** 5. Review Comments to the Author Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters) Reviewer #1: The paper presents an approach to detecting scenarios of data drifting in PM2.5 levels using various strategies, and based on these findings, proposes a CNN attention-based transfer learning model for forecasting PM2.5 levels. This approach signifies an important step towards addressing the challenges posed by data drift in environmental data analysis. However, while the exploration of such ideas is intriguing and potentially impactful, there are several concerns that obscure the paper's contribution. Primarily, the abstract does not adequately reflect the methodological proposition involving CNNs with attention mechanisms and training using an alternative loss function. Instead, it focuses predominantly on the detection techniques for data drifting and their comparison with other state-of-the-art models in time series forecasting. This omission raises questions about the thoroughness with which the proposed models are presented and their potential advantages over existing methods. It is not clear if the integration of advanced neural network architectures and novel loss functions for improved accuracy in PM2.5 prediction is also a paper's contribution as the title states. Clarify. It's also pertinent to note that the paper's engagement with the state of the art in time series forecasting for PM2.5 prediction does not explicitly mention Long Short-Term Memory (LSTM) networks with attention mechanisms. This omission could suggest that such approaches have not been widely proposed or explored within this specific domain, marking a potential gap in the literature that the current study aims to address. The introduction of CNNs with attention mechanisms tailored for PM2.5 forecasting, as highlighted in the paper, is an innovative step. However, the absence of a discussion on LSTM networks with attention mechanisms raises questions about the comprehensive coverage of existing methodologies and their comparative analysis within the study. Furthermore, the paper references work number 33 in the state of the art section on transfer learning, which also proposes CNNs with attention mechanisms for forecasting tasks. This reference necessitates a clear delineation of how the current paper's approach diverges from the methodologies previously outlined. The distinction appears to lie solely in the specific focus on data shifting strategies and the implementation of a novel loss function. Please Clarify in the text. In terms of the presentation of results, the paper's Figure 1 lacks clarity due to the absence of units on the y-axis. To enhance the readability and usefulness of this figure, it is advisable to not only include units but also consider revising the representation of time. Instead of using the week number of the year, incorporating reference dates that can easily pinpoint the months covered would significantly improve the ability to identify potential seasonal patterns within the data. Such a modification would make the figure more intuitive and facilitate a deeper understanding of the temporal dynamics being illustrated. Regarding Figure 2, which presents a comparison of the time series distributions, there is a notable absence of quantitative insights into the specific differences observed between the distributions. While visual comparisons can be informative, they leave much to the reader's interpretation and assumptions about the nature and significance of the differences. To address this, the paper should include a detailed analysis of the distributions, highlighting significant quantitative differences. For Figures 3 to 6, where the results are presented, it's critical to address the issue of readability and interpretation posed by the extensive use of grouped bars. The dense clustering of bars makes it challenging to discern patterns or draw meaningful insights from the data. The authors are advised to consider alternative visualization techniques that can facilitate a clearer understanding of the results. One such recommendation is the use of heat maps to represent the average values, and/or the standard deviations of the different metrics found across all measurement points. Heat maps could provide a visually intuitive means to identify variations and patterns in the data, highlighting areas of interest more effectively than the current bar groupings. Another suggestion is to employ a matrix format that includes references to the years and represents the distribution shapes of the metrics found (e.g., probability functions or boxplots). This approach would allow for a more nuanced comparison across different years combinations, making it easier to identify where metrics are generally higher or lower. Furthermore, it's essential that all figures include titles on their axes. For the presentation of time series forecasting results, it's indeed beneficial to go beyond the use of Root Mean Square Error (RMSE) as the sole metric of accuracy. Including comparisons using Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) is recommended, as these metrics are also widely utilized in the field and provide different perspectives on forecast accuracy. Lastly, it's paramount to ensure clarity and consistency in the notation used throughout the paper, especially in the mathematical equations presented. Every symbol, variable, and function within these equations should be explicitly defined and described within the text. Reviewer #2: The authors have worked on Predicting PM2.5 levels by using an Enhanced Deep neural network; and discussed the Data drift problem along with Wrapped Loss Function. Their findings have shown the validity of their work. But, to improve the paper to a level of acceptance, the following major revisions are required: 1. The last sentence of the abstract is not complete. 2. For the readers’ advantage, they need to elaborate on the point description of the proposed FLC and BLC models along with better and more comprehensive/professional-looking diagrams. In its present form, the paper looks like copy pasted from a scientific report or a thesis. 3. In the title they mention the use of Deep NN, but in the model description they mention CNN. Why they have not mentioned CNN in the title? 4. In the Model Design Section (FLC & BLC), they mentioned the conclusion along with a description (although inadequate) of the models (in lines 320-322: The results clearly demonstrate that both the front-loaded and back-loaded models proposed outperform other models, and the wrapped model exhibits superior performance compared to the one without the wrapped loss function). Is it required here? 5. A complete description of the datasets created/used should be given in its section before the Experimental setup. 6. The results of the Experiments should be illustrated in a separate Results Section. 7. Results should be discussed appropriately in a separate Discussion Section. 8. The Conclusions should be drawn as per the findings in the research work and they should be substantiated with the outcomes. 9. The organization to present the paper needs to be improved. They may read the papers to improve the presentation. https://doi.org/10.3390/diagnostics12051134 https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9740199 https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9956807 At the end of the Introduction section, describe the organization of the paper in brief. 10. The purpose of using many statistical methods should be properly explained in the paper. 11. A comparison of the models with SOTA-published works should be given. 12. The related Work/literature review section should be more elaborate to show the substantial research gaps and accordingly the research problem statement should be mentioned after the Introduction section. ****** 6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy. Reviewer #1: No Reviewer #2: No ******** [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step. https://doi.org/10.1371/journal.pone.0314327.r001
Revision 1
8 May 2024 Author Response Academic Editors Comment 1: (general comment) Please ensure that your manuscript meets PLOS ONE’s style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals. plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_ title_authors_affiliations.pdf Response I have followed the the above format in my manuscript Comment ) 2:Please note that PLOS ONE has specific guidelines on code sharing for submissions in which author generated code underpins the findings in the manuscript. In these cases, all author-generated code must be made available without restrictions upon publication of the work. Please review our guidelines at https://journals.plos.org/plosone/s/materials-and-software-sharingloc-sharing-code and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse. Response Definitely, I will share the source code and so that it can reproduce. Comment 3) Thank you for stating the following financial disclosure: "This document is the result of the research project funded by the National Science and Technology Council of Taiwan, under the grant NSC 109-2119-M-001-010-A." Please state what role the funders took in the study. If the funders had no role, please state: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript." If this statement is not correct you must amend it as needed. Please include this amended Role of Funder statement in your cover letter; we will change the online submission form on your behalf. Response This study was supported by National Science and Technology Council of Taiwan, under the grant NSC 109-2119-M-001-010-A. No additional external funding was received for this study. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Comment 4) Specific Comment 4: Please note that funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form. Please remove any funding-related text from the manuscript. Response: I already removed from the Acknowledge sections. I already removed from the Acknowledge sections. Comment 5) Specific Comment 5: When completing the data availability statement of the submission form, you indicated that you would make your data available on acceptance. We strongly recommend all authors decide on a data sharing plan before acceptance, as the process can be lengthy and hold up publication timelines. Please note that, though access restrictions are acceptable now, your entire data will need to be made freely accessible if your manuscript is accepted for publication. This policy applies to all data except where public deposition would breach compliance with the protocol approved by your research ethics board. If you are unable to adhere to our open data policy, please kindly revise your statement to explain your reasoning and we will seek the editor’s input on an exemption. Please be assured that, once you have provided your new statement, the assessment of your exemption will not hold up the peer review process. Response: The data we collected are called EPA data and shared in the below link https://1drv.ms/f/c/ 615854b8e839ca3e/EpO3bYopjRBHjEWAWlO0WtIBU0b2NPiGliOhrsdKMxRCwA?e=RfX5z7 and after preprocessing the EPA data collected only for Taipei area and the evaluation process used this datasets https://1drv.ms/f/c/615854b8e839ca3e/EvywgMKXuJRLian31QPWT0IBVSCWtKVWc9fu4vrqaNEKCQ? e=txnbOZ Comment 6. Please amend either the abstract on the online submission form (via Edit Submission) or the abstract in the manuscript so that they are identical. Response :I have edited the abstract from the online submission. Comment 7. Please ensure that you refer to Figure 12 in your text as, if accepted, production will need this reference to link the reader to the figure. Response I have confirmed about figure 12 and it’s included in the manuscript. Comment 8. We note you have included a table to which you do not refer in the text of your manuscript. Please ensure that you refer to Tables 4 and 8 in your text; if accepted, production will need this reference to link the reader to the Table. Response: I have confirmed Tables 4 and 8 so that their text is included in the manuscript. Reviewer 1 Comment Reviewer 1: (general comment) This omission raises questions about the thoroughness with which the proposed models are presented and their potential advantages over existing methods. It is not clear if the integration of advanced neural network architectures and novel loss functions for improved accuracy in PM2.5 prediction is also a paper’s contribution as the title states. Clarify. Response I have discussed in detail about contributions and clarified the contributions in the end of the Abstract Section. In this work, the proposed model and the loss function are both contributions for this work to yanking the data drifting scenario. The CNN-Attention-based model solves the data drifting issue, and the wrap loss function also helps alleviate the data drifting problem with model training and works for the baseline model and neural network models to achieve more correct results. Comment 1) Specific Comment 1: However, the absence of a discussion on LSTM networks with attention mechanisms raises questions about the comprehensive coverage of existing methodologies and their comparative analysis within the study. Response After getting feedback from the reviewers, it is added the LSTM attention-based model and included in the Model Designs section. It found that their result could not compete with our proposed model. The following sentence is added to the Results and Discussion section. We have discussed LSTM attention with their loss functions too. In the LSTM-attention model is also applied to the time series [43] prediction, in our work we have applied for hourly PM2.5 prediction its results are included and compared with other models. Comment 2) Specific Comment 2: Furthermore, the paper references work number 33 in the state of the art section on transfer learning, which also proposes CNNs with attention mechanisms for forecasting tasks. This reference necessitates a clear delineation of how the current paper’s approach diverges from the methodologies previously outlined. The distinction appears to lie solely in the specific focus on data-shifting strategies and the implementation of a novel loss function. Please Clarify in the text. Response Line 33, I have added our work necessities and compared them with the work that has been used as a reference outlined in [33]. They trained their model using data from the last four years and used it to predict the PM2.5 level of Taipei City for the next year without addressing the data drifting issue. Taking advantage of abundant data to tackle data drifting challenges through transfer learning, they introduced a CNN attention-based model designed to make hourly predictions from time series data, 144 thereby improving the prediction performance and overcoming domain adaptation 145 challenges specific to time series data. This work focused on solving the problem where the new city developed and did not have enough data, and in some cases, they did not have enough pollutant observation of pollutant sources in the area. The authors transferred the knowledge from one enriched city Beijing to another city Baoding. They needed the source city and a target city for transfer learning, but in our proposed model, it does not require any source or target city. Also, they did not discuss the data drifting issues. Our proposed model has one more advantage that it predicts from EPA data and weather data from the last couple of years and also includes domain adaptation Comment 3) Specific Comment 3: In terms of the presentation of results, the paper’s Figure 1 lacks clarity due to the absence of units on the y-axis. To enhance the readability and usefulness of this figure, it is advisable to not only include units but also consider revising the representation of time. Instead of using the week number of the year, incorporating reference dates that can easily pinpoint the months covered would significantly improve the ability to identify potential seasonal patterns within the data. Such a modification would make the figure more intuitive and facilitate a deeper understanding of the temporal dynamics being illustrated. Response: I have added the figure which I included dates with the month. I have added the in the figure 1 according to the reviewer’s comments. it is stated in line 23, Introduction Section In Fig 1 states the monthly change with specific dates can easily be visible where drifts are available. Fig 1. Weekly Average PM2.5 Levels of Years from 2014-2017 Comment 4) Specific Comment 4: Regarding Figure 2, which presents a comparison of the time series distributions, there is a notable absence of quantitative insights into the specific differences observed between the distributions. While visual comparisons can be informative, they leave much to the reader’s interpretation and assumptions about the nature and significance of the differences. To address this, the paper should include a detailed analysis of the distributions, highlighting significant quantitative differences. Response I have explained the data distribution difference from the figure 2. It is discussed in line 35-39 Fig 2 shows the probability distributions of PM2.5 levels over various years, highlighting the noticeable differences between them. As the annual phenomenon of PM2.5 levels is evident, our study adopts an annual perspective as a basis to observe and investigate the prediction of PM2.5 levels. Fig 2 shows the normal distributions of PM2.5 levels over various years, highlighting the noticeable differences between them. In Figure 2 it is shown that the yearly histogram can also check the yearly comparison since the annual phenomenon of PM2.5 levels is evident, our study adopts an annual perspective as a basis for observation and investigation of the prediction of PM2.5 levels. Fig 2. PM2.5 distributions from Year 2014 to 2017 Comment 5) Specific Comment 5: For Figures 3 to 6, where the results are presented, it’s critical to address the issue of readability and interpretation posed by the extensive use of grouped bars. The dense clustering of bars makes it challenging to discern patterns or draw meaningful insights from the data. The authors are advised to consider alternative visualization techniques that can facilitate a clearer understanding of the results. One such recommendation is the use of heat maps to represent the average values, and/or the standard deviations of the different metrics found across all measurement points. Heat maps could provide a visually intuitive means to identify variations and patterns in the data, highlighting areas of interest more effectively than the current bar groupings. Another suggestion is to employ a matrix format that includes references to the years and represents the distribution shapes of the metrics found (e.g., probability functions or boxplots). This approach would allow for a more nuanced comparison across different years combinations, making it easier to identify where metrics are generally higher or lower. Furthermore, it’s essential that all figures include titles on their axes. Response After getting the feedback from the reviewers, I have added the heat maps for each methodology and it has been shown from Fig 3-4 to Fig 4-5 and Fig 5-6 . In-Line included numbers are 253-254,267-268,277-278. I have added the Heat map figure so changes can be easily identified from the different statistical techniques Fig 3. Yearly comparisons using JS divergence for all stations Fig 4. Yearly comparisons using JS divergence for four stations by heatmap In Fig 4 includes the highest four stations P value by JS divergence, it is presented by heatmap diagram so can clearly understand the value differences, this value is calculated in the time of yearly comparison, here in Datong and Guting has maximum values. Fig 5. Yearly comparisons using Pearson Correaltion Fig 6. Yearly comparisons using PC for four stations by heatmap In Fig 6 included the highest four stations of P value by MMD, it is presented by heatmap diagram so can clearly understand the value differences, here Wanli and Xizhi stations has maximum P value. Fig 7. Yearly comparisons using MMD for all stations Fig 8. Yearly comparisons using MMD for four stations by heatmap In Fig ?? included the highest four stations P value by PC, it is presented by heatmap diagram so can clearly understand the value differences, Datong and Wanli has maximum values. Comment 6) Specific Comment 6: For the presentation of time series forecasting results, it’s indeed beneficial to go beyond the use of Root Mean Square Error (RMSE) as the sole metric of accuracy. Including comparisons using Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) is recommended, as these metrics are also widely utilized in the field and provide different perspectives on forecast accuracy. Response Thank you for your advice. After getting the advice of the reviewers, I have added the RMSE, MAE, and MAPE results for PM2.5 predictions. In the Evalution Criterian section I discussed its definition. And the Results and Discussion section I discussed about its MAE and MAPE results. To evaluate the performance of the baseline models and the proposed model, we have applied different methods such as the root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) to evaluate their performance. We have shown their results in Results Section. Comment 7) Specific Comment 7: Lastly, it’s paramount to ensure clarity and consistency in the notation used throughout the paper, especially in the mathematical equations presented. Every symbol, variable, and function within these equations should be explicitly defined and described within the text. 8Response I have carefully checked the consistency of the notation used in this paper, and also the equations. Reviewer 2: Comment Reviewer 2: (general comment) 1. The last sentence of the abstract is not complete. Response There was a mistake. I have completed the first line. Additionally, we introduced a wrapped loss function incorporated into a model, resulting in more accurate results compared to those using the original loss function alone. Comment Reviewer 2: (general comment) For the readers’ advantage, they need to elaborate on the point description of the proposed FLC and BLC models along with better and more comprehensive/professional looking diagrams. In its present form, the paper looks like copy pasted from a scientific report or a thesis. Response After getting feedback from the reviewer, I have redrawn it following the advice. It is redrawn in the section in Model Designs in the subsection in FLC and BLC models. Fig 10. Front-loaded connection model Fig 11. Back-loaded connection model Comment Specific Comment 2-1: In the title they mention the use of Deep NN, but in the model description they mention CNN. Why they have not mentioned CNN in the title? Response: I have edited my title according to the feedback from reviewers. On the Data Shifting: Predicting PM2.5 levels by using Enhanced Deep neural network enhanced based nural networkOn the Data Shifting: Predicting PM2.5 levels using CNN-based neural network Comment Specific Comment 2-2: In the Model Design Section (FLC and BLC), they mentioned the conclusion along with a description (although inadequate) of the models (in lines 320-322: The results clearly demonstrate that both the front-loaded and back-loaded models proposed outperform other models, and the wrapped model exhibits superior performance compared to the one without the wrapped loss function). Is it required here? Response: Attachments Attachment Submitted filename: Response to Reviewers.pdf https://doi.org/10.1371/journal.pone.0314327.r002
23 Aug 2024 Decision Letter - Worradorn Phairuang, Editor PONE-D-24-05738R1On the Data Shifting: Predicting PM2.5 levels by using CNN based neural networkPLOS ONE Dear Dr. Hossen, Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. ============================== ACADEMIC EDITOR: Major revisions ============================== Please submit your revised manuscript by Oct 07 2024 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. Please include the following items when submitting your revised manuscript: A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols. We look forward to receiving your revised manuscript. Kind regards, Worradorn Phairuang, Ph.D. Academic Editor PLOS ONE [Note: HTML markup is below. Please do not edit.] Reviewers' comments: Reviewer's Responses to Questions Comments to the Author 1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation. Reviewer #3: (No Response) Reviewer #4: All comments have been addressed ******** 2. Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented. Reviewer #3: Partly Reviewer #4: Yes ****** 3. Has the statistical analysis been performed appropriately and rigorously? Reviewer #3: No Reviewer #4: Yes ****** 4. Have the authors made all data underlying the findings in their manuscript fully available? The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified. Reviewer #3: (No Response) Reviewer #4: No ****** 5. Is the manuscript presented in an intelligible fashion and written in standard English? PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here. Reviewer #3: No Reviewer #4: Yes ****** 6. Review Comments to the Author Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters) Reviewer #3: The paper still needs significant improvement because the current version does not clearly explain the proposed methods and results. My comments for improvement as follows, 1. Abstract is not clear about what the paper proposes. 2. The paper needs to clearly outline the main contributions are, such as analysis of data drifting, drift detection, deep learning model (FLC and BLC Models), or the application of wrapped loss function in the models. 3. The introduction should clearly and explicitly describe the problem statement and the contributions. 4. In Previous work section, the paper frequently uses the subject "authors". It is unclear whether this refers to the authors of the current paper or the authors cited in the references. 5. In lines 112-119 in page 4, the statement "Among them, CNN-based models stand out for their 115 ability to process historical data from EPA stations based on the spatial distribution of 116 the sites [28]" attempts to conclude that CNN is better based on [28]. Among various models for various data in refs [22-28], it would be better to clearly explain why CNN is preferred, supported by strong evidence. 6.The Previous work section does not discuss the attention-base model which is proposed by the paper. 7. From Fig 3 to Fig 8, it is unclear what the y-axis values represent. Are they the results from Eq (3), (4), (5) and (6) or just P-value? It is important to note that the interpretations of the results from PC differ those from other methods. 8. In page 12, it is better explain FLC and BLC models more detail. 9. It is strongly recommended to reorganize the paper. Reviewer #4: (No Response) ****** 7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy. Reviewer #3: No Reviewer #4: Yes: Dr. Nishit Aman ******** [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step. Attachments Attachment Submitted filename: reviewer_comments.docx https://doi.org/10.1371/journal.pone.0314327.r003
Revision 2
1 Sep 2024 Author Response Plos Reviewers Response Comment 1: Abstract is not clear about what the paper proposes. Response After getting the feedback,I have written it, and elaborated on the purposes of my work. Comment 2. The paper needs to clearly outline the main contributions are, such as analysis of data drifting, drift detection, deep learning model (FLC and BLC Models), or the application of wrapped loss function in the models. Response I have written the main contributions of the paper in the introduction section Comment 3. The introduction should clearly and explicitly describe the problem statement and the contributions. Response I have rewritten the problem statement and contribution clearly in the Introduction section. Comment 4. In Previous work section, the paper frequently uses the subject "authors". It is unclear whether this refers to the authors of the current paper or the authors cited in the references. Response I have edited the authors from the paper and stated clearly with the author’s name of the paper. Comment 5. In lines 112-119 in page 4, the statement "Among them, CNN-based models stand out for their 115 ability to process historical data from EPA stations based on the spatial distribution of 116 the sites [28]" attempts to conclude that CNN is better based on [28]. Among various models for various data in refs [22-28], it would be better to clearly explain why CNN is preferred, supported by strong evidence. Response After getting feedback, I have included a few works where that applied CNN based model for PM2.5 predictions the lines are given below from 119-124. Comment 6.The Previous work section does not discuss the attention-base model which is proposed by the paper. . Response I have discussed the attention base model on the previous work section. Comment 7. From Fig 3 to Fig 8, it is unclear what the y-axis values represent. Are they the results from Eq (3), (4), (5), and (6) or just P-value? It is important to note that the interpretations of the results from PC differ from those of other methods. From Fig 3 to Fig 8 the results are from equations 3-6, those equations calculated the P value finally. The interpretation of those results is given there. Response From the equation (3) Eq (3), (4), (5), and (6), calculated the P value and value are illustrated on Figure 3-8. Heatmaps are based on the P value that we get after those equations are calculated. Comment 8. In page 12, it is better to explain FLC and BLC models in more detail. Response The model description, I have described those models in more detail in FLC and BLC models sections. Comment 9. It is strongly recommended to reorganize the paper. Response I organize the paper according to feedback. 11. Plos second Reviewers Response Comment o Line 98-203: I think there is no need of this Section giving too much theoretical information. It could be more appropriate for a review paper. It is better to shorten this part and include as part of the introduction. Response I have edited the section and shortened it as I needed to discuss the previous work so I kept it and the another reviewer also asked to introduce the applied model in details.. Comment o Add a section on Study Area and Data. Give details about the study area with Map with location of various monitoring stations. Response I have added a section in the name Study area and data and drawn a map on the location and described it. Comment o Include a Section on Methods giving full details in sequence about the method used in the study. A typical machine learning related work should include section as Data collection and pre-processing, EDA and Feature engineering, Model development and evaluation, and finally prediction. Response I have discussed the data collection and analysis of how collected the data in Data collection and pre-processing section . There are several section as Model development and prediction . Comment o . In figures, the default design for axes, labels, and legend has been used. Make label bold and enlarge. Response I have redrawn the figures and made them bold too accordingly. Comment o Write which software packages or libraries were used for the statistical analysis. Give references too. Response I have given the references for the work’s statistical analysis as there are three techniques so I gave all of them references too. Comment o In conclusion, add a few sentences on policy implications and future work. Add limitations here. Response I have added future work and limitations too in the following paragraph. Attachments Attachment Submitted filename: Response to Reviewers.pdf https://doi.org/10.1371/journal.pone.0314327.r004
1 Nov 2024 Decision Letter - Kristofer Lasko, Editor PONE-D-24-05738R2On the Data Shifting: Predicting PM2.5 levels by using CNN based neural networkPLOS ONE Dear Dr. Hossen, Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. ============================== ACADEMIC EDITOR: In an effort to provide a speedy and rigorous review, I am your new academic editor. Accordingly, I have reviewed your manuscript and the revisions provided by the reviewers. Beyond what the reviewers have already said, I would like for a few minor changes to be added to ensure your manuscript is easily understood, and located by the appropriate audience. Specifically, Please ensure your figures have sufficient description. The description for most figures is too short. You will also want to provide units for PM2.5, which I could not locate in figure captions or in many parts of the text. (i.e. MAE is in the units of PM2.5 concentration, but units are not listed). consider a revision to your manuscript title. As it is, the title does not inform the reader about all of the major work that you've completed and is not representative of the manuscript. Current title: "On the Data Shifting: Predicting PM2.5 levels by using CNN based neural network". My critique is that "data shifting" is too vague and you've done more than just use CNNs. You should revise the title, a potential recommendation could be something like "Improving PM2.5 pollution predictions by correcting for annual data drift with wrapped loss functions evaluated against CNNs and LSTMs". Please feel free to use the suggested or modify as you like. Please check over the text for readability and English in the manuscript. For example, the use of "yanking" in the abstract is confusing and that sentence should be reworked. Consider adding specific major metrics in your conclusion and abstract sections (i.e. how much did your wrapped loss function improve on the accuracy when dealing with data drift?); this will increase audience interest and usability of your findings. Ensure abbreviations and terminology are consistent throughout the manuscript (i.e. p-value vs P value). Lastly, keywords. Considering adding more if allowed. Additionally, consider related terms beyond abbreviations. Upon completion, I will review your article as soon as it comes back in and provide you with a timely final decision. I will not send it back to reviewers. ============================== Please submit your revised manuscript by Dec 18 2024 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. Please include the following items when submitting your revised manuscript: A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols. We look forward to receiving your revised manuscript. Kind regards, Kristofer Lasko, PhD Academic Editor PLOS ONE Journal Requirements: Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice. [Note: HTML markup is below. Please do not edit.] Reviewers' comments: Reviewer's Responses to Questions Comments to the Author 1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation. Reviewer #4: All comments have been addressed Reviewer #5: All comments have been addressed ******** 2. Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented. Reviewer #4: Yes Reviewer #5: Yes ****** 3. Has the statistical analysis been performed appropriately and rigorously? Reviewer #4: Yes Reviewer #5: Yes ****** 4. Have the authors made all data underlying the findings in their manuscript fully available? The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified. Reviewer #4: Yes Reviewer #5: Yes ****** 5. Is the manuscript presented in an intelligible fashion and written in standard English? PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here. Reviewer #4: Yes Reviewer #5: Yes ****** 6. Review Comments to the Author Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters) Reviewer #4: (No Response) Reviewer #5: Two models that consider the characteristics of data drifting for PM2.5 prediction are proposed and all comments have been addressed. ****** 7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy. Reviewer #4: Yes: Dr. Nishit Aman Reviewer #5: No ******** [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step. https://doi.org/10.1371/journal.pone.0314327.r005
Revision 3
7 Nov 2024 Author Response Point-by-Point Responses and the Reviewer Comments of "Enhancing PM2.5 Prediction by Mitigating Annual Data Drift Using Wrapped Loss and Neural Networks" by Md Khalid Hossen, Yan-Tsung Peng, Meng Chang Chen Comment 1: Please ensure your figures have sufficient description. The description for most figures is too short. You will also want to provide units for PM2.5, which I could not locate in figure captions or in many parts of the text. (i.e. MAE is in the units of PM2.5 concentration, but units are not listed) Response After getting the feedback, I have included the units for MAE (μg/m3), PM2.5(μg/m3) and unit for RMSE (μg/m3) too. I have included in the text and in their figure too. Comment 2. consider a revision to your manuscript title. As it is, the title does not inform the reader about all of the major work that you’ve completed and is not representative of the manuscript. Current title: "On the Data Shifting: Predicting PM2.5 levels by using CNN based neural network". My critique is that "data shifting" is too vague and you’ve done more than just use CNNs. You should revise the title, a potential recommendation could be something like "Improving PM2.5 pollution predictions by correcting for annual data drift with wrapped loss functions evaluated against CNNs and LSTMs". Please feel free to use the suggested or modify as you like. Response Thank you for your suggestion and I modify a little .The current tile are given below "Enhancing PM2.5 Prediction by Mitigating Annual Data Drift Using Wrapped Loss and Neural Networks". Comment 3. Please check over the text for readability and English in the manuscript. For example, the use of "yanking" in the abstract is confusing and that sentence should be reworked. Response I have revoked the sentence. I have checked the text and readibilty and modified accordingly. Comment 4. Consider adding specific major metrics in your conclusion and abstract sections (i.e. how much did your wrapped loss function improve on the accuracy when dealing with data drift?); this will increase audience interest and usability of your findings. Response Thank you for your suggestions. I have included the model performance percentage how much they improved it and included in the Abstract and conclusion. Comment 5. Ensure abbreviations and terminology are consistent throughout the manuscript (i.e. p-value vs P value). Response After getting feedback, I have fixed the manuscript only P value . Comment 6. Lastly, keywords. Considering adding more if allowed. Additionally, consider related terms beyond abbreviations. Response I have added more keyword . Attachments Attachment Submitted filename: Response to Reviewers.pdf* https://doi.org/10.1371/journal.pone.0314327.r006
11 Nov 2024 Decision Letter - Kristofer Lasko, Editor Enhancing PM2.5 Prediction by Mitigating Annual Data Drift Using Wrapped Loss and Neural Networks PONE-D-24-05738R3 Dear Dr. Hossen, We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements. Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication. An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. If you have any questions relating to publication charges, please contact our Author Billing department directly at authorbilling@plos.org. If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org. Kind regards, Kristofer Lasko, PhD Academic Editor PLOS ONE Additional Editor Comments (optional): Reviewers' comments: https://doi.org/10.1371/journal.pone.0314327.r007
Formally Accepted
18 Nov 2024 Acceptance Letter - Kristofer Lasko, Editor PONE-D-24-05738R3 PLOS ONE Dear Dr. Hossen, I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now being handed over to our production team. At this stage, our production department will prepare your paper for publication. This includes ensuring the following: * All references, tables, and figures are properly cited * All relevant supporting information is included in the manuscript submission, * There are no issues that prevent the paper from being properly typeset If revisions are needed, the production department will contact you directly to resolve them. If no revisions are needed, you will receive an email when the publication date has been set. At this time, we do not offer pre-publication proofs to authors during production of the accepted work. Please keep in mind that we are working through a large volume of accepted articles, so please give us a few weeks to review your paper and let you know the next and final steps. Lastly, if your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org. If we can help with anything else, please email us at customercare@plos.org. Thank you for submitting your work to PLOS ONE and supporting open access. Kind regards, PLOS ONE Editorial Office Staff on behalf of Dr. Kristofer Lasko Academic Editor PLOS ONE https://doi.org/10.1371/journal.pone.0314327.r008

Open letter on the publication of peer review reports

PLOS recognizes the benefits of transparency in the peer review process. Therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. Reviewers remain anonymous, unless they choose to reveal their names.

We encourage other journals to join us in this initiative. We hope that our action inspires the community, including researchers, research funders, and research institutions, to recognize the benefits of published peer review reports for all parts of the research system.

Learn more at ASAPbio .