A Hybrid Forecast Model of EEMD-CNN-ILSTM for Crude Oil Futures Price

Round 1

Reviewer 1 Report

Please see the attached file.

Comments for author File: Comments.pdf

Overall, the language is good. However, some revisions are needed.

Disclaimer: I am not a native English speaker.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This manuscript aims to propose a hybrid model called EEMD-CNN-ILSTM for crude oil futures. This paper uses the crude oil futures data of the Shanghai Energy Exchange in China as the experimental data set. The results of the experiment show the model is more effective and accurate. Based on my review, the manuscript has several comments that necessary to be considered. The following comments may help enhancing the quality of this manuscript.

 

 

General comments:

[1].   Using a complex model based on integrating the ensemble empirical mode decomposition (EEMD), Convolutional Neural Network (CNN), and Improved Long Short-Term Memory (ILSTM) is a complex choice to identify the crude oil futures. Because using complex models require several hyperparameters that need to be carefully considered with several assumptions. In the same time, there are many simple models, e.g, CNN-LSTM can be used for such propose without the need to try complex models. The authors need to present the different between these methods and the proposed technique

[2].   The authors are recommended to pay attention to the use of tense in the text. For example, in abstract, past tense should be used when describing the work already done. Please review and revise accordingly.

[3].   The abstract should be based on the principles of writing a correct and reasonable abstract. In the abstract, the authors need to have distinct three parts: i) what topic they are going to investigate, ii) which method and assumptions are implemented, and iii) finally, what are the important results.

[4].  No key contributions from this study can be found in introduction section. What is the new in this hybrid model? How did the authors identify the crude oil futures, Uncertainties of estimation are not provided? Can this hybrid model be applied to any other case project and what is the relevant limitations? Typically, authors should present the new in this research and the different between this research and current studies. Otherwise, the presented reference can guide you to present the recent machine and deep learning techniques, e.g., DOI: 10.1016/j.tust.2023.105104

[5].   The CNN-ILSTM model was integrated with the EEMD-technique to identify the crude oil futures. What are other feasible alternatives? What are the advantages of adopting this soft computing technique over others in this case? How will this affect the results? The authors should provide answer with details on this.

 

[6].   This study aims to predict the crude oil futures based on hybrid DL model. However, what are the source of information that used in this study? What are the relevant sensitivity for these conditions? What are the sources of these data? How did the authors apply these data in this hybrid EEMD-CNN-ILSTM model?

 

[7].   X and y axis as well as the relevant title, need to be added in the figures 1, 2, 13, 14. Otherwise, several equations have been indicated without the relevant references.

 

[8].   Authors need to present the model reliability based on Friedman analysis or any other technique and determine the model efficiency compared to other state of art models. More information related to Friedman analysis can be found in https://doi.org/10.1007/s11440-022-01461-4.

 

[9].   In the conclusion section, the limitations of this study, suggested improvements of this work and future directions should be highlighted.

 

technical writing can be improved 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

A hybrid forecast model of EEMD-CNN-ILSTM for crude oil futures price is proposed. The experiments show the model is more effective and accurate than the six prediction models.  

The four components of EEMD need to be justified. The authors stated that "the ZCR of IMF1 is higher than 40% as a high-frequency component, the ZCR of IMF2  and IMF3 is between 10% and 40% as medium-frequency components, and the ZCR of IMF4 ~ IMF9 is less than 10% as low-frequency components. Therefore, the number of the prediction model’s input data components is reduced from ten to a fixed number of four." It is not clear why the frequency components are classified into four categories. For example, what if more than four categories are used? Any evidence showing that the four categories are better than other alternatives?   What makes the crude oil forecast unique compared with other sequence forecast tasks? Please provide more explanation on the crude oil data characteristics which could justify the choice of  EEMD-CNN-ILSTM. 

Acceptable. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

·        L63. Please pay attention on the punctuation

·        L287-289: there is nothing wrong here. However, the conventional notation for any model is using y as dependent variable, and x as predictors.

The reading is quite smooth right now.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

no comments

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

N/A

N/A

Author Response

Please see the attachment.

Author Response File: Author Response.docx