Tourism Demand Forecasting Based on a Hybrid Temporal Neural Network Model for Sustainable Tourism

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Abstract

1. Clarity and Accessibility: The abstract is concise but overly technical, making it less accessible to a broader audience. For clarity, terms such as BiLSTM and ARIMA should be briefly explained or simplified.

2. Practical Implications: The abstract needs to adequately emphasize the practical significance of the findings for sustainable tourism. Highlighting specific applications, such as implications for policymakers or tourism managers, would enhance its relevance and appeal.

Introduction

1. Literature Context: While the introduction cites some foundational references, it lacks engagement with recent and relevant studies from 2023–2024. Including these would strengthen the context and demonstrate the study’s relevance to ongoing research.

2. Research Contribution: The manuscript needs to explicitly articulate how the proposed hybrid model advances the field of sustainable tourism forecasting. Clearly state the theoretical or practical gap the study aims to address.

3. Structure: Conclude the introduction with an outline of the paper’s structure to improve the reader’s navigation and understanding of the manuscript.

Research Methodology and Model Construction

1. Methodology Justification: While the BiLSTM-Transformer model is described in detail, the rationale for selecting this hybrid approach over alternatives must be sufficiently justified. Elaborate on why this combination is particularly suitable for tourism demand forecasting.

2. Data Preprocessing: The steps are outlined but need references to established methods or guidelines. Provide citations and explain the significance of these methods in enhancing data quality.

3. Hyperparameter Details: More details about the hyperparameter configurations (e.g., number of layers, dropout rates, batch sizes) and the training process are needed to enhance reproducibility.

Experimental Results and Analysis

1. Comparative Analysis: While the manuscript demonstrates that the hybrid model outperforms other approaches, it needs to discuss why this model achieves superior performance sufficiently. Provide a deeper analysis of the hybrid model’s advantages in handling specific challenges in tourism demand forecasting.

2. Real-World Implications: The discussion focuses on technical performance metrics but fails to explore practical implications for tourism management and policy-making. Discuss how the findings can inform sustainable tourism practices or decision-making processes.

 Conclusions

1. Contributions: The conclusions reiterate the results but fail to highlight the study’s theoretical or practical contributions to sustainable tourism forecasting. Emphasize these contributions clearly.

2. Limitations and Future Research: The limitations section is underdeveloped. Discuss potential limitations of the model (e.g., generalizability to other contexts or datasets) and suggest concrete directions for future research.

3. Practical Applications: Expand on how tourism stakeholders, such as policymakers and operators, can use the proposed model to make data-driven decisions. Highlight its potential for improving resource allocation and policy development.

References

1. Quantity and Relevance: The number of references is insufficient to support the study’s claims. Incorporate more recent studies to strengthen the manuscript’s foundation and ensure academic rigor.

2. Balance: Ensure a balanced mix of foundational studies and the latest developments in hybrid modeling and sustainable tourism forecasting to better contextualize the research.

Specific Suggestions

1. Short Biography of Authors: This section is not part of the standard manuscript structure for Sustainability and should be removed to align with the journal’s guidelines.

2. Visual Elements: If Figures A, B, and C do not contribute meaningfully to the analysis, they should be removed to streamline the paper and enhance its focus.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

1. Technical Language: Simplify technical terms and avoid excessive jargon to improve readability and accessibility. For instance, briefly explain the “self-attention mechanism” more straightforwardly.

2. Clarity: Revise sentences to improve clarity and conciseness. Some sections are verbose, while others need more explanation or detail.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The article attempts to develop a model to forecast tourism demand and enhance the management of tourism flows, a timely and relevant topic in the pursuit of sustainable tourism practices.

While, the methodological section is highly detailed and meticulously explained, the authors fail to adequately contextualize their research within existing studies. This omission makes it difficult to understand how their model builds upon or diverges from prior research on tourism demand measurement and management. A stronger engagement with extant literature would enhance the article’s scholarly contribution.

The conclusions section is underdeveloped in terms of practical applications. Although the proposed model shows potential, the authors do not sufficiently explain how policymakers, tourism managers, or industry stakeholders could apply these findings to improve decision-making and manage tourism flows effectively.

Recommendations for Improvement:

• Incorporate a more extensive review of prior research on tourism demand forecasting and management. Highlight gaps in the literature that the current study addresses.
• Clearly outline the implications of the findings for policymakers and industry practitioners. Include actionable strategies for leveraging the model in sustainable tourism planning.

Addressing these areas would significantly enhance the article’s contribution to the field.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The authors proposed a hybrid time-series neural network model for forecasting tourism demand. The results of the study hold significant implications for achieving sustainable tourism development. However, the following issues are present in this paper:

（1）This paper employs models to predict sustainable tourism demand. Although the authors present existing models, the review section on tourism demand forecasting studies that utilize these models lacks sufficient detail. A more thorough review of demand forecasting studies employing these models is recommended.

（2）References are cited in an inconsistent format, e.g., 32 lines “Lawetal.(2019)”, “and output gates[? ] “. Citation formatting must be accurate and consistent.

(3) Section 2.1, Data Description and Preprocessing, is too brief. It is recommended that a more comprehensive description of the data context be included. Additionally, the detailed process of how the data was processed should be clearly outlined, allowing readers to replicate the procedures described in the article and verify its rigor.

(4) The discussion section of the manuscript is insufficient. A more comprehensive comparative analysis should be conducted between the results of this study and the existing literature to emphasize the innovative aspects of the research.

（5）The literature citation is insufficient. The authors should conduct a more thorough review of the relevant literature, particularly in the area of study, to enhance the theoretical foundation of the paper and increase its academic rigor.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

Manuscript ID: sustainability-3390004-peer-review-v1

Thank you for the opportunity to review this paper entitled “Tourism Demand Forecasting Based on a Hybrid Temporal Neural Network Model for Sustainable Tourism”. I recommend a major revision if the author(s) can improve the paper for further consideration. Kindly consider my comments as preliminary and subject to further updates. Here are the reasons behind my decision:

 

Abstract

• The abstract could be clearer about the novelty of the hybrid model itself. While it mentions the BiLSTM and Transformer components, it does not explicitly explain how combining these two methods leads to a unique contribution or what challenges this hybrid approach addresses.
• Consider adding a sentence that explicitly states the gap in the existing literature or methods that this hybrid model fills (e.g., why a hybrid model is better suited to this problem than using BiLSTM or Transformer alone).
• The phrase “complex time series data” is somewhat vague. It would be helpful to clarify what aspects of the tourism demand data make it complex (e.g., seasonality, trends, external factors like economic events, etc.).
• The term "outperforms" is used, but a little more detail on why the hybrid model performs better could be helpful. Does it outperform in certain scenarios or consistently across all conditions?
• The abstract mentions the model's utility for “sustainable policy formulation and resource management in the tourism sector.” While this is a good point, a brief elaboration on how the model directly contributes to these areas would be helpful. For example, does it help forecast demand during peak seasons, reduce over-tourism, or optimize resource allocation? Providing a direct link between the model's forecasting capability and sustainable tourism outcomes would make this statement stronger.

·         The keywords are relevant, but you may consider expanding them to include terms like "sustainability," "deep learning," and perhaps "hybrid model" to better capture the essence of the paper and improve discoverability.

Introduction

• Although the introduction discusses the importance of accurate forecasting and mentions various models, it could more clearly articulate the specific gap that this study is addressing. For example, it mentions that hybrid models are relevant but does not specify why combining BiLSTM and Transformer in particular is the right approach to tackle the challenges of sustainable tourism forecasting. A more explicit statement about what makes this hybrid combination novel or necessary would strengthen the argument.
• Consider including a brief sentence that directly connects the limitations of existing methods (e.g., ARIMA, standalone BiLSTM, or standalone Transformer) in the context of tourism demand forecasting and explains how this hybrid model overcomes those limitations.
• While the introduction mentions the influence of holidays and events (e.g., Chinese New Year, Songkran Festival) on tourism demand, it could emphasize more how these factors pose specific challenges for forecasting. For example, the introduction could describe how traditional forecasting methods might struggle with accurately predicting demand during such irregular events.
• While the sustainability angle is mentioned, the connection between the hybrid model and sustainable tourism could be elaborated upon further. How exactly does the forecasting model contribute to sustainability (e.g., reducing over-tourism, enabling more precise resource allocation)?
• The introduction cites various sources but could better integrate these references within the text to enhance the flow and readability. For instance, the references to Law et al. (2019), UNWTO, and other studies are somewhat scattered and could be placed more cohesively within the narrative. This would strengthen the logical progression of the argument.

Research Methodology and Model Construction

• While the dataset is well-described, it would be beneficial to include more specific information about the total number of records, regions covered, and frequency of data collection (e.g., daily, weekly, monthly). This would help readers gauge the dataset's complexity and its potential challenges for modeling.
• Additionally, a brief mention of the data’s limitations, such as potential biases (e.g., underreporting in certain regions), could provide a more balanced view.
• Outlier detection is mentioned, but a brief explanation of how outliers were identified and corrected would be helpful. For example, were statistical techniques like z-scores or interquartile range (IQR) used?
• The attention formula (Equation 3) is introduced but lacks clarity in terms of how it’s applied in the tourism demand forecasting context. It would be beneficial to explain how this formula relates to the specific task and what role each of the variables (Q, K, V) plays in the Transformer architecture. While this equation is widely known, explaining how it contributes to capturing long-range dependencies in tourism demand would connect theory to application.
• While the ARIMA model’s limitations are discussed well, a little more detail on how it might still be effective in simpler scenarios (for instance, when the data is relatively stationary or when there’s minimal seasonal variation) would present a more balanced view. This would also help highlight the relative strengths of the deep learning models when applied to complex, non-linear time series data.

 Discussion

• This part is absent. It should be added. Authors must pay a great attention in this part on how exactly the findings contribute to the theory and overall body of knowledge. What new information authors providing to industry professionals that they are not already aware of?

 

Implications

• This part is absent. It should be added. Authors must pay a great attention in this part on what is the practical contribution of the study? Authors have to enrich the manuscript with specific and more implications to practice. Providing a clearer link between the research findings and actionable recommendations would strengthen the practical relevance.

 

Further research avenues

• This part is absent. I do recommend adding this part.

 

Language

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

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Abstract: Simplify technical terminology and emphasize the study’s practical applications, particularly for policymakers and tourism managers.

Introduction: Include references to recent studies (2023-2024) to enhance relevance. Clearly articulate the research gap and contribution of the hybrid BiLSTM-Transformer model.

Methodology: Provide a more substantial justification for the chosen model and add more details on hyperparameter configurations and training processes to ensure reproducibility.

Results and Analysis: Discuss why the hybrid model performs better than alternatives. Connect findings to real-world applications in tourism management.

Conclusions: Expand on the study’s contributions to sustainable tourism forecasting. Address limitations and propose concrete future research directions.

References: Add a comprehensive reference list with recent and relevant studies to support the study’s claims and enhance academic rigor.

Comments on the Quality of English Language

Improve clarity and flow by simplifying technical terms and restructuring certain sections for better readability.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The revised manuscript does not include a list of references, making it impossible to determine whether the citations are appropriate. The revised manuscript does not include a list of references, making it impossible to determine whether the citations are appropriate. It is essential to ensure that all cited works are included in the reference list and that all references listed are cited within the text.

Author Response

Comments 1: The revised manuscript does not include a list of references, making it impossible to determine whether the citations are appropriate. The revised manuscript does not include a list of references, making it impossible to determine whether the citations are appropriate. It is essential to ensure that all cited works are included in the reference list and that all references listed are cited within the text.

Response 1: Thank you for pointing out this oversight. We sincerely apologize for the omission of the reference list in the previously submitted version. This was an unintentional error during the file preparation process. We have now included the complete and correctly formatted reference list in the newly revised manuscript. We have double-checked to ensure all cited works are included in the reference list and that all references listed are cited within the text. We believe this addresses your concern regarding the appropriateness of our citations.

Reviewer 4 Report

Comments and Suggestions for Authors

Thanks for improvements you have made on the manuscript.

Author Response

Comments 1: Thanks for improvements you have made on the manuscript.

Response 1: Thank you for your positive feedback and for acknowledging the improvements made to the manuscript. We appreciate your time and effort in reviewing our work. We are glad that the revisions have addressed your previous concerns.

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript needs significant revisions to align with the reviewer's comments. Specific efforts are required in the areas of:

1. Simplifying and clarifying technical terms.
2. Expanding on practical implications for tourism stakeholders.
3. Engaging with recent literature.
4. Enhancing methodological justifications and details.
5. Providing actionable insights and practical applications in the conclusions.

Comments on the Quality of English Language

Simplify technical language, clarify practical implications, and ensure grammatical accuracy.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf