Methods for Mid-Term Forecasting of Crop Export and Production
Round 1
Reviewer 1 Report
This paper presents neural network models for mid-term forecasting of crop production and export. The topic is interesting and the authors resolve an important problem using machine learning methods. However, the paper must be improved. My comments are as follows:
1) Authors must definitely improve the related work section and they must cite more papers. Thus, the motivation of the paper will be highlighted.
2) Authors must clarify the method section. Also, what parameters have been used in the methods?
3) Are authors planning to share the dataset with the research community? Because the combination of this dataset can help researchers to find the combination of various datasets easily.
Author Response
1) Authors must definitely improve the related work section and they must cite more papers. Thus, the motivation of the paper will be highlighted.
The related work section is revised (lines 124-151). We've considered six additional papers and added discussion regarding the drawbacks of the existing approaches (NARX, LSTM) for the mid and long-term forecasting.
2) Authors must clarify the method section. Also, what parameters have been used in the methods?
The method section has been completely revised. We've added more details in Sections 3.2.1 and 3.2.2, especially, regarding the values of the hyperparameters (lines 224, 228-232, 267-269, etc.).
3) Are authors planning to share the dataset with the research community? Because the combination of this dataset can help researchers to find the combination of various datasets easily
We've added the link to the dataset to Section 3.1 (line 200)
Reviewer 2 Report
The paper presents an interesting analysis of different neural networks models that might be useful for mid-term forecasting of wheat production and export. Overall, the paper is well written and provides methodologically sound results. Nevertheless, there are some issues that should be addressed:
Mayor issues:
- Introduction [page 1, line 29]. Which recent sanctions? Sanctions related to Russia or something else?
- Literature is missing. The facts stated in the Introduction would be sound if concrete literature is cited. This is especially referring to paragraph #1 and #2 (lines 54-57).
- Clear added value of the paper is missing. What is this paper adding to the existing literature?
- What are the disadvantages of the NARX? Advantages are stated on page 3 (lines 116-118).
- Figure 2. Focus is on models used for forecasting a virus pandemic impact. Nevertheless, this is not in a focus of this paper. Please revise.
- Using Twitter messages for the analysis. The issue is which tweets are considered: i) using tweets from the relevant institutions would have higher weights as people involved in wheat production and trade might intentionally follow these institutions. Tweets from individuals might have lower weight; ii) The methodological problem is that only tweets in English could be considered. Nevertheless, the relevance of the tweets in national language is crucial. Overall, these methodological issues should be clearly stated as limitations.
Minor issues:
- Abstract [page 1, line 11]. Please change the word order to: "The mid and long-term predictions...";
- Abstract [page 1, line 20]. Add "." at the end of the sentence.
- Introduction [page 1, line 31]: around instead of other?
- Introduction [page 2, line 50]: ":" at the end of the sentence.
- Introduction [page 2]. I would recommend to switch the order of the last two paragraphs.
Author Response
1. Introduction [page 1, line 29]. Which recent sanctions? Sanctions related to Russia or something else?
Related to Russia. But, we've removed that mention of sanctions because that would make the results country-specific:
The COVID-19 pandemic triggered a drop in all types of production, a decrease in trade flows, and a rupture of cross-country communications all around the World [1].
2. Literature is missing. The facts stated in the Introduction would be sound if concrete literature is cited. This is especially referring to paragraph #1 and #2 (lines 54-57).
Two references are added (lines 56, 57).
-
- Clear added value of the paper is missing. What is this paper adding to the existing literature?
We've added the value for practical applications (lines 65-68).
The proposed models have value in practical applications since they provide accurate mid-term forecasting of crop production and export, and consider non-structural features, which can significantly affect the markets.
3. What are the disadvantages of the NARX? Advantages are stated on page 3 (lines 116-118).
We've added discussion regarding the advantages and disadvantages of the NARX and LSTM in the Related Work section. (lines 127-134, 145-147).
4. Figure 2. Focus is on models used for forecasting a virus pandemic impact. Nevertheless, this is not in a focus of this paper. Please revise.
The figure and the whole section have been revised. Additional information regarding the network's hyperparameters and training process is provided.
5. Using Twitter messages for the analysis. The issue is which tweets are considered: i) using tweets from the relevant institutions would have higher weights as people involved in wheat production and trade might intentionally follow these institutions. Tweets from individuals might have lower weight; ii) The methodological problem is that only tweets in English could be considered. Nevertheless, the relevance of the tweets in national language is crucial. Overall, these methodological issues should be clearly stated as limitations
Added as the limitations at the end of Section 3.2.3. (lines 324-328)
The described methodology has the following limitations. We do not distinguish between tweets from the relevant institutions and messages from individuals. Besides, we consider only tweets in English, which limits the completeness of the results. In the future, the official messages shall have a higher weight than the individuals. We are also going to utilize a cross-lingual pipeline, which would help us to collect that information.
Minor issues:
- Abstract [page 1, line 11]. Please change the word order to: "The mid and long-term predictions...";
- Abstract [page 1, line 20]. Add "." at the end of the sentence.
- Introduction [page 1, line 31]: around instead of other?
- Introduction [page 2, line 50]: ":" at the end of the sentence.
- Introduction [page 2]. I would recommend to switch the order of the last two paragraphs.
All fixed, as it is proposed.
Round 2
Reviewer 1 Report
The authors improved the paper based on my comments and feedback. It is now ready for publication.
Reviewer 2 Report
I have no further comments and suggestions. All of my previous comments are addressed in the improved version of the paper.
