Ensemble Empirical Mode Decomposition Granger Causality Test Dynamic Graph Attention Transformer Network: Integrating Transformer and Graph Neural Network Models for Multi-Sensor Cross-Temporal Granularity Water Demand Forecasting

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This work presents a novel framework for forecasting water demand based on the integration of Graph Neural Networks and Transformer Network, both of which are data-driven models. The paper is well-structured, with a well-written introduction that clearly states the novelties and presents the results effectively. The work conducted is quite impressive. Setting a p-value of 0.01 demonstrates the reliability of the research outcomes.

I have a few comments:

The quality of the figures needs improvement.
Please support lines 26-27 with the following reference: “Burst detection in water distribution systems: The issue of dataset collection.”
Future Work/Discussion: Could the authors comment on the potential coupling between EG-DGATN and physics-based models (see, for example, “The extension of EPANET source code to simulate unsteady flow in water distribution networks with variable head tanks”)? I believe integrating a water distribution network (WDN) model to support the sensor network could significantly enhance the forecasting capabilities of EG-DGATN, especially if synthetic cases related to bursts and leakages are simulated using the WDN model.
Could you provide some data regarding the computational performance of the presented framework?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

he paper introduces a novel approach called Ensemble Empirical Mode Decomposition Granger causality test Dynamic Graph Attention Transformer Network (EG-DGATN) for water demand forecasting. This method combines Transformer and Graph Neural Networks, using the EEMD-Granger test to analyze sensor interconnections. It extracts spatio-temporal features through dynamical graph spatio-temporal attention layers. Experimental results show that EG-DGATN outperforms baseline models, achieving improvements in MAPE metrics and an impressive R2 score of 0.97, indicating high predictive accuracy and explanatory power. The proposed model demonstrates potential applications in predictive tasks for smart water management systems. However, the paper requires substantial revision to elucidate the scientific objective and clarify the results and conclusions.There are several key points that need clarification:

1-The introduction section could benefit from clearer statements and explanations, particularly regarding the research focus, conceptual/theoretical framework, and relevant literature review studies informing the current research.

2-What is the main objective of the paper regarding water demand forecasting?

3-How does the EG-DGATN approach differ from traditional forecasting models?

4-What role does Ensemble Empirical Mode Decomposition (EEMD) play in the proposed method?

5-How does the Granger causality test contribute to understanding sensor interconnections?

6-What are the key components of the EG-DGATN model, particularly in terms of combining Transformer and Graph Neural Networks?

7-How does the model address spatio-temporal features in the causal domain?

8-What were the baseline models used for comparison, and how did EG-DGATN outperform them?

9-What are the specific improvements in MAPE metrics observed at different forecasting intervals?

10-How does the R2 score of 0.97 contribute to evaluating the model's performance?

11-In what ways does the proposed EG-DGATN model exhibit potential applications in smart water management systems?

12-Enhance the conclusion section by including more detailed results.

13-Discuss the applicability of your findings and propose directions for future studies in the field.

14-Recommend language editing to improve the overall clarity and coherence of the manuscript.

15-Suggest grammar editing to ensure the manuscript adheres to standard grammatical conventions

Comments on the Quality of English Language

Moderate editing of English language required

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have satisfactorily addressed the issues that I have raised in my initial review and therefore recommend acceptance of the revised manuscript without further revision.