All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
I confirm that the authors have addressed the reviewers' comments.
[# PeerJ Staff Note - this decision was reviewed and approved by Mehmet Cunkas, a PeerJ Section Editor covering this Section #]
no comment
no comment
no comment
Please see the reviewers' detailed comments. The paper contains critical discrepancies in comparative metrics (e.g., misstated benchmark returns). Strengths include a clear methodology, comprehensive performance metrics, and domain relevance. However, improvements are needed in result consistency, terminology, and expanded analysis (e.g., explainability via SHAP/LIME, error metrics, and future scope for transformers/web apps). Suggested enhancements include tabular result comparisons, pseudocode, a "Contribution and Motivation" section, and a "Comparison with Literature" subsection to bolster contribution and validity.
**PeerJ Staff Note:** Please ensure that all review, editorial, and staff comments are addressed in a response letter and that any edits or clarifications mentioned in the letter are also inserted into the revised manuscript where appropriate.
**Language Note:** PeerJ staff have identified that the English language needs to be improved. When you prepare your next revision, please either (i) have a colleague who is proficient in English and familiar with the subject matter review your manuscript, or (ii) contact a professional editing service to review your manuscript. PeerJ can provide language editing services - you can contact us at [email protected] for pricing (be sure to provide your manuscript number and title). – PeerJ Staff
**PeerJ Staff Note:** It is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors agree that they are relevant and useful.
The paper is well written, but it still needs some improvements
Convert Related Work text to table
Write an algorithm of the approach in pseudocode
Also create table of results which will show the numeric values clearly.
Also add subsection of “Contribution and Motivation” after introduction
Add Error Analysis in the paper to add more value
Add Accuracy, precision, recall, f1 score, auc, roc etc. in the paper. It will add value in the results section.
Also use SHAP and LIME in the paper. It will add more value in the paper
Also add Future Scope after conclusion
Also add one section of “Comparison with existing literature” before conclusion section. It helps to add novelty in the research.
Also mention novelty in your paper
Strengths:
1. Clear Problem Statement and Innovation: The article clearly identifies the core challenges in agricultural financial transactions (noise, time-series dependence, limited adaptability) and proposes specific innovations (Grubbs-MAD denoising, LSTM-enhanced DRL, transaction cost-based reward function, rolling training, multi-factor features) to address them.
2. Strong Quantitative Results: The abstract and results sections present compelling quantitative evidence of the model's performance, including annualized return (45.12%), reduction in maximum retracement (35%), and Sharpe ratio (1.51, 62% improvement). This provides a strong basis for the claims of superiority.
3. Well-Defined Methodology: The experimental setup, including data sources, stock selection, and the phased rolling training strategy, is clearly articulated, enhancing the reproducibility of the study.
4. Effective Denoising Strategy: The detailed explanation and experimental validation of the Grubbs-MAD denoising method demonstrate its significant positive impact on model stability and profitability, highlighting the importance of data preprocessing.
5. Leveraging LSTM for Time-Series Data: The paper effectively argues for and demonstrates the benefits of integrating LSTM networks to capture long-term dependencies in time-series financial data, a crucial aspect for market prediction.
6. Comprehensive Performance Metrics: The use of final asset value, profitability, Sharpe ratio, AARR, and MDD provides a holistic view of the model's financial performance and risk management capabilities.
7. Discussion of Micro and Macroeconomic Impacts: The discussion section effectively extrapolates the model's benefits to both individual agricultural stakeholders (farmers, enterprises) and broader macroeconomic goals (food security, resource sustainability, policy insights), showcasing the practical relevance and potential impact.
8. Acknowledgement of Limitations and Future Work: The conclusion thoughtfully outlines current limitations (data dependence, adaptability, computational complexity, lack of domain-specific knowledge integration) and suggests promising avenues for future research (attention mechanisms, transformers, scalability, real-time deployment).
9. Domain-Specific Relevance: The focus on agricultural financial transactions is a novel and important application area for DRL, making the research highly relevant to smart agriculture initiatives.
Improvements Required:
1. Critical Discrepancy in Comparative Results (Section 4.3): This is the most significant issue.
o The statement that the proposed model's 45.12% annualized return "is substantially higher than the A3C-LSTM model[24] at 83.51% and the DDPG-LSTM model[25] at 89.93%" is a direct contradiction. 45.12% is lower than 83.51% and 89.93%. This must be corrected immediately. Either the numbers are wrong, or the interpretation is wrong, or the benchmark models' results are misattributed/misunderstood. This casts serious doubt on the claimed superiority in this specific comparison.
o Clarify the "annualized return of the DDPG-LSTM decision model in this paper is 56.32%". Is this a variation of the proposed model, or a separate comparison? This contradicts the 45.12% stated for the main model in the abstract and conclusion, and the 89.93% for DDPG-LSTM mentioned just lines before. Ensure consistent naming and clear delineation of results for the proposed model vs. benchmarks.
2. Refine Terminology Consistency:
o Ensure consistent use of "LSTM-DQL" or "LSTM-DQN" throughout if they refer to the same proposed model. The abstract uses "LSTM-DQL" while the discussion often refers to "LSTM-DQN architecture." While DQL and DQN are closely related, precision helps.
3. Minor Phrasing and Conciseness:
o "A multi-factor dynamic denoising framework is developed by combining the Grubbs test with the median absolute deviation (MAD) method, classifying agricultural financial indicators into six feature types to enhance noise robustness." – This sentence is a bit long. Consider splitting it for better readability
4. Are you planning any explainability in your model to make black box to transparent model atleast add it in future scope as 1 or 2 sentences about how to make your model transparent. Check the below papers or add any article which is discussing about XAI
Enhancing Transparency in Smart Farming: Local Explanations for Crop Recommendations Using LIME
Role of Explainable AI in Crop Recommendation Technique of Smart Farming
Streamlit-based enhancing crop recommendation systems with advanced explainable artificial intelligence for smart farming.
5. Whether deep learning techniques like transformers encoder decoder suitable for your research. consider below paper or any other paper. atleast consider adding deep learning techniques in future scope. Analysis of An Intellectual Mechanism of a Novel Crop Recommendation System using Improved Heuristic Algorithm-based Attention and Cascaded Deep Learning Network
6. Whether any web application designed for user access , atleast consider in future scope for web application design, check below for reference
AI-Driven Smart Farming Portal: Overcoming Language Barriers and Enhancing Agricultural Productivity through Machine Learning
Software application to prevent suicides of farmers with asp. net MVC
no comment
no comment
no comment
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.