Short- and Medium-Term Electricity Consumption Forecasting Using Prophet and GRU

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1) The introduction provides a good overview of related works. However, it could be enhanced by providing a bit more context on why short- and medium-term electricity consumption forecasting is crucial and how it impacts energy system operations.

2) In Section 2, elaborate more on the GRU and Prophet models, including their working principles and strengths, to give readers a clear understanding before proposing their combination.

3) Provide details about the preprocessing steps applied to the collected data. Explain any data normalization, scaling, or outlier handling performed, and justify these choices.

4) In Section 4, elaborate on the experimental setup, including the division of the dataset into training and testing, parameter tuning methods, and evaluation metrics used. Specifically, mention if cross-validation was applied.

Author Response

Dear reviewer

We extend our sincere appreciation to the reviewers for their valuable feedback and constructive suggestions. We have carefully reviewed and implemented the recommended changes, which are indicated in the manuscript in blue color (P and L refer to page and line numbers). Your input has significantly enhanced the quality and clarity of our article. We are grateful for your contributions to our work.

Best regards,
Namrye Son.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This study introduces a novel approach for electricity consumption forecasting. The authors integrate a DNN-based recurrent model, well-suited for long-term forecasting, with the Prophet model, which is capable of handling seasonality and events. I think that the paper is interesting and tackles a high-impact problem within the forecasting community. Nevertheless, in its current state, the manuscript exhibits the following limitations:

 

1) The manuscript exhibits inadequate writing quality, with numerous language errors and mistakes dispersed throughout the text. These issues hinder the comprehension of the authors' arguments and conclusions. I think the manuscript necessitates a thorough language review, including the revision of numerous sentences (e.g., the enumeration in the abstract in not needed), as well as enhancements to the figures to enhance overall readability.

 

2) The paper ignores recent literature on load/electricity forecasting techniques. Demand time series in smart grids and power systems can be naturally arranged in a hierarchical structure, and these time series can be forecasted using hierarchical forecasting models. For instance, load demand data may be correlated (e.g. the demand of a given region may be close to that of a neighboring region) and connected to each other through aggregation constraints (e.g. the consumption at the grid level is the sum of each region’s consumption). Recent papers suggest that hierarchical forecasting models are competitive with sophisticated deep-learning approaches. See these papers:

- https://doi.org/10.1016/j.eswa.2021.115102

- https://doi.org/10.1016/j.ijforecast.2021.05.011

 

3) I feel that the validation phase is difficult to reproduce. It is crucial that the authors, in addition to providing results, give the reader the tools to verify that these results are correct as well as scientifically sound. Although the experimental setup is clear, the configuration of GRU is missing (e.g., number of layers, activation function, units, etc.).

Comments on the Quality of English Language

-

Author Response

Dear reviewer

We extend our sincere appreciation to the reviewers for their valuable feedback and constructive suggestions. We have carefully reviewed and implemented the recommended changes, which are indicated in the manuscript in blue color (P and L refer to page and line numbers). Your input has significantly enhanced the quality and clarity of our article. We are grateful for your contributions to our work.

Best regards,
Namrye Son.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

In this article, the author introduces an innovative approach to electricity consumption prediction, offering better results in terms of accuracy compared to the conventional Prophet model.

Please see my comments below:

Row 20: Please clarify “more than 23 times”. How to get this value, e.g., using which evaluation metric?

Figure 2: y-axis is energy consumption, kWh? In figure 2, there are two blue lines: dark and light. What is the difference between these two blue lines? Why does the predicted value stop in 2019-06?

 Row: 233: How many observations are in the final model development data? 31 days in June?

Row 237: Training data collected from July 2018 to May 2019 is not consistent with July 1, 2018, to June 31, 2019 mentioned in row 232.

Row 345: Table 2 shows the testing results. What are the training results? Need to add more information about the training part.

Author Response

Dear reviewer

We extend our sincere appreciation to the reviewers for their valuable feedback and constructive suggestions. We have carefully reviewed and implemented the recommended changes, which are indicated in the manuscript in blue color (P and L refer to page and line numbers). Your input has significantly enhanced the quality and clarity of our article. We are grateful for your contributions to our work.

Best regards,
Namrye Son.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Accept in present form.

Author Response

Dear Reviewer,

Thank you for reviewing our paper. Your valuable feedback has greatly contributed to the improvement of our paper's quality. We appreciate your cooperation.

Best regards,

Namrye Son.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have effectively addressed all my concerns, resulting in an enhanced paper quality. However, they must conduct a meticulous review of the bibliography due to a few errors. The paper incorrectly references the newly added articles [23, 24, 25], where it seems that the first and last names of each author have been swapped. Please rectify this mistake and provide the correct reference information. The right references are as follows:

[23] Mancuso, P., Piccialli, V., & Sudoso, A. M. (2021). A machine learning approach for forecasting hierarchical time series. Expert Systems with Applications, 182, 115102.

[24] Brégère, M., & Huard, M. (2022). Online hierarchical forecasting for power consumption data. International Journal of Forecasting, 38(1), 339-351.

[25] Shi, X., Chen, Z., Wang, H., Yeung, D. Y., Wong, W. K., & Woo, W. C. (2015). Convolutional LSTM network: A machine learning approach for precipitation nowcasting. Advances in Neural Information Processing Systems, 28.

Please also ensure that all the remaining references in the paper are thoroughly reviewed and checked for accuracy. 

Comments on the Quality of English Language

-

Author Response

Dear Reviewer,

Thank you for your thorough review of our paper, and we appreciate your positive feedback regarding the improvements made. We are committed to enhancing the quality of our work.

We have attached the revised paper, based on the points you mentioned.

Thank you for your cooperation and helfness.

Best kindness,

Namrye Son.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

I think the authors have answered my questions in a good way.

Author Response

Dear Reviewer,

Thank you for reviewing our paper. Your valuable feedback has greatly contributed to the improvement of our paper's quality. We appreciate your cooperation.

Best regards,

Namrye Son.