A Novel Hybrid Model for Financial Forecasting Based on CEEMDAN-SE and ARIMA-CNN-LSTM

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

 

1. Please carefully polish and clean up all figures and tables. All figures and tables in a paper must be consistent. Please make sure all figures are high-resolution (300 dpi)

2. To enhance the paper's impact, the author should discuss potential future directions for this research.

3. Provide additional explanations of several parameters used, including RMSE, MAE, and MAPE. Is there standardization of the parameters used?

4. in line 325, it says, "According to the information in the figure, the range of sample entropy values for  IMF1-IMF3 is similar, with the sample entropy values for the first three components fluctuating between 0.0307 and 0.0328.". please explain what is entropy and what is the standardization of the entropy value

5. in the section methodology, add algorithms from the proposed model to clarify how the proposed model works

6. In line 375, it says " The DW value is 1.9966, which is closer to 2, indicating that there is no significant autocorrelation among the residual series which means the residual series are independent".  Explain the dw value, and what is the standardization of this value.

7. Please add the explanation about Figures 3 and 4.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Since the ARIMA models are weak in non-linear financial time series models, the researcher had to adopt the traditional ARCH and GARCH models and compare them with deep learning models.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

Dear Authors,

I believe this manuscript has its own merits. have several comments as below. I hope you will correct along with most of my comments.

 

1. Clarity and Focus

The manuscript is generally well-written and covers a complex topic with clarity. The flow of the paper is logical, and the integration of various methodologies is explained in a structured manner. However, there are sections where technical jargon is dense, which might be challenging for readers who are not experts in the field.

 

2. Figures and Tables

The figures and tables provided are helpful in understanding the model's performance and the experimental results. However, some figures (e.g., flowcharts) could be enhanced with clearer labels and descriptions to improve readability.

 

The introduction provides a good overview of the challenges in financial time series prediction and the relevance of the proposed method. Add more recent references to highlight the novelty of your approach including below references.

 

Lee, Y., & Kim, E. (2024). Deep Learning-based Delinquent Taxpayer Prediction: A Scientific Administrative Approach. KSII Transactions on Internet & Information Systems, 18(1).

 

Heo, W., Kim, E., Kwak, E. J., & Grable, J. E. (2024). Identifying Hidden Factors Associated with Household Emergency Fund Holdings: A Machine Learning Application. Mathematics, 12(2), 182.

 

3. Methodology

The explanation of the CEEMDAN algorithm and its integration with ARIMA and CNN-LSTM is thorough. However, a brief comparison with other decomposition methods could strengthen the justification for choosing CEEMDAN.

The process flowchart in Figure 2 is useful but could benefit from more detailed annotations to guide the reader through each step of the hybrid model.

 

4. Experimental Setup

The experimental setup section is well-detailed. It would be beneficial to include more information on the choice of hyperparameters for the CNN-LSTM model and any tuning strategies employed. Consider providing more context on the dataset used, such as the time span of the data and any preprocessing steps taken before analysis.

 

5. Results and Discussion

The results are presented clearly, and the comparative analysis with other models is insightful.

 

6. Conclusion

The conclusion succinctly summarizes the key findings and implications of the study. Adding a few sentences on potential real-world applications of your model could enhance the impact of your work. Highlighting specific avenues for future research, such as exploring other types of financial data or incorporating additional machine learning techniques, would provide a more comprehensive outlook.

Comments on the Quality of English Language

I want you to review and check your own manuscript for minor grammatical errors and ensure that the language is precise and formal.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The author has made improvements and revisions according to the reviewer's request.

Reviewer 3 Report

Comments and Suggestions for Authors

Dear Authors,

I have checked the revised version. It seems now bettern than the earlier version. Especially, the conclusion part is pretty much improved. Great work.