Using Machine Learning Methods to Predict Consumer Confidence from Search Engine Data
Round 1
Reviewer 1 Report
This manuscript deals with machine learning methods to predict consumer confidence from search engine data. This thesis deals with social science content. However, the contents of the CS field, such as CNN, deal with contents that are far from the latest contents.
For this reason, this paper is recommended for consideration after major revision.
Comment
In Research Method,
1. The author describes CNN using the 2009 definition. Modern CNNs work by updating errors using the backpropagation algorithm. In addition to the definition asserted by the author, it is necessary to add an up-to-date definition.
2. On page 4, line 161, I want to ask if the author thinks the polling layer is essential.
3. On page 5, math formula (1), please rewrite the formula neatly. Case and subscript distinction are messed up.
Typos (checking please):
1. On Abstract, line 14: in the internet à on the internet
2. P2, line 68: Random Forest Regresssion à Random Forest Regression
3. P2, line 69: Convolutional Neural Network ,à Convolutional Neural Network, (deleting blank)
4. P3, line 90: the paradigm of economics research à the paradigm of economics research, (comma)
5. P3, line 111: keywords Google Trends à keywords in Google Trends
6. P4, line 144: socio-economic à socioeconomic
Author Response
I am very grateful for your comment, which helps me to improve my research a lot. I have upload the document of the response to the comment. Thanks very much. Merry Christmas!
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The language of the paper must be reviewed, and very long sentences (without using full stops) are presented so that comprehensibility is lost. There are double spaces, unclear sentences, use of poor English language rules.
The state-of-the-art review should be extended with more related works.
The research method proposed in this work exposes CNN as a deep learning method (Section 3.1) to carry out the prediction; however, the fundamentals of the machine learning methods still need to be exposed. In addition, finally, a machine learning method was better, as mentioned in the abstract.
An analysis of why a machine learning method turned out better than deep learning methods is needed when the advances in Artificial Intelligence are in the direction of deep learning.
It needs to be clear how searching data using keywords to predict Consumer Confidence Index is achieved. Does it remain to explain what processing is done on the data: natural language processing techniques?, feature extraction?
The results presented in the paper need to be clarified. For example, “errors of various machine learning and deep learning models” is not explained, nor what the values in Tables 4 and 5 mean.
The performance of the methods is evaluated with the MAE and RMSE metrics, which are not exposed or explained: what range does it handle? What would be an encouraging value?
The work presents a contribution to the task of predicting the effect of the Consumer Confidence Index. However, using deep learning or machine learning methods in Artificial Intelligence does not present an original contribution.
The citations throughout the text need to be in the correct format, and they should be reviewed. Reference 16 has a different font.
Author Response
I am very grateful for your comment, which helps me to improve my research a lot. I have upload the document of the response to the comment. Thanks very much. Merry Christmas!
Author Response File: Author Response.pdf
Reviewer 3 Report
The topic of the paper seems to be interesting in the context of the now-a-days economy.
The connection with the literature and the highlight of the gap should be further improved as it is not clear to which extent the results of the paper can be used in practice.
Furthermore, the abstract should be restructured as there is a lot of text referring to methodology and fewer referring to the results. Please discuss more the results and less the methodology.
The methodology is not clear. Please better explain it. Also, it is not clear how the 64 keywords have been selected and how the authors have ensured that the search has resulted in results that are connected to the CCI. Please add references for the selected keywords.
Furthermore, how the influence of the noisy data has been excluded?
Please add more information related to the comparison of the models considered in Table 4.
Please pay attention to the typos along the paper (e.g. in the abstract: Cndex is used instead of Index).
Please also pay attention to the references - in the paper there are included only the names of the authors, while the year the paper has been published is missing. Please check the format of the references requested by the journal.
Author Response
I am very grateful for your comment, which helps me to improve my research a lot. I have upload the document of the response to the comment. Thanks very much. Merry Christmas!
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Through the revision, this paper has been improved. However, it still needs some refinement in terms of preparation. For this reason, I recommend this paper for publication after minor revision.
Comment
In Response,
1. The page the author mentioned is wrong.
Line 226-238 à 209-221
Line 221 à 209 (needing line alignment)
In line 209(math formula), can you replace the upper sum k with n. Because the upper sum k and variable k are duplicated.
Typos (checking please):
1. In line 46, Chinacould à China could
2. L 86, It promote à It promotes
3. L 90, Macine à Machine
4. L 234, so as to à to
----- Continued. -----
Author Response
Thanks for your comments, which helps me improve my research writing a lot, I am very grateful for it. Here is my revision:
Comment:
- Lines 209-221 are the text, and line 222 is the secondary heading. In line 209-222, I have modified the format according to the regulations.
- In line 205(math formula), I have already replaced the upper sum k with n.
Typos:
- In line 46, I have already changed “Chinacould” to “China could”.
- In line 86, I have already changed “It promote” to “It promotes”.
- In line 90, I have already changed “Macine” to “Machine”.
- In line 235, I have already changed “so as to” to “to”.
Other small modifications:
- In addition, in line 10, I changed the correspondence’s email address.
- In line 225, I have already changed “It will” to “It will”(delete one space).
If I still have problems with my manuscript, feel free to let me know and I will try my best to revise it, thank you very much.
Reviewer 2 Report
The comments have been taken care of. The paper has improved significantly.
Author Response
Thank you very much for your comments, my research writing has improved a lot and I will work harder on my future paper research and writing.
Reviewer 3 Report
Thank you for the revised version of the paper. I have no further comments.
Author Response
Thank you very much for your comments, my research writing has improved a lot and I will work harder on my future paper research and writing.
Round 3
Reviewer 1 Report
In the 2nd round, the correction was not made properly, but in the 3rd round, the comments were well made. For this reason, I recommend this paper in 'Accept in present form'.
