Prediction of Losses Due to Dust in PV Using Hybrid LSTM-KNN Algorithm: The Case of Saruhanlı

Round 1

Reviewer 1 Report

The document lacks contextualization in the introduction and literature review; the contributions are not clearly outlined. The methodology is overly simplistic and lacks a clear rationale for the experimental design. The charts, figures, and other visuals fail to effectively illustrate the evaluated concepts or results. The Hybrid LSTM-KNN algorithm is not clearly explained in terms of its application and utility. Finally, the results and discussion section is deficient in relevant information, diminishing the overall validity of the document.

Author Response

Thank you.

Author Response File: Author Response.pdf

Reviewer 2 Report

The following comments are provided to improve the quality of the article.

- the Turnitin software indicate a similarity of 22 %.

- the topic is very interesting, but the information is very difficult to follow and understand; please find a method of presenting the information through comparative analysis in the form of a table – based of references, so that the reader can understand the information more easily.

- the prediction of losses used algorithms can also be applied to other areas (different SEM) or localities/countries? 

- in Conclusion section, authors should detail beside the original research contributions of the paper, positive aspects, observed deficiencies and suggestions on how to improve them.

Author Response

Thank you.

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper investigates a hybrid deep learning model to predict losses caused by dust in PV panels installed in Manisa Saruhanlı district. This topic is interesting, and the structure is complete, some minor revisions are needed before acceptance, e.g.,

1)The word like innovative is meaningless, which can be deleted.

2)The reference citation is not right, e.g., [9] should be placed after Kouz et al, i.e., Kouz et al [9], the same to others.

3)How to predict dust losses in PV panels using the hybrid LSTM-KNN algorithm? This should be clarified.

4) Figs,3, 6, 8, 9, 11 and 14 blur, and clearer ones can be used, and the text in the figures can be larger and Times New Roman format to make it clearer.

5) The conclusions part can be incorperated into one paragraph.

6) What's the difference between predictor and observer in your viewpoint?

none

Author Response

Thank you.

Author Response File: Author Response.docx

Reviewer 4 Report

A hybrid deep learning model called LSTM-KNN was used to predict losses caused by dust in PV panels in the paper. Additionally, sensitivity analysis of the input parameters was conducted through the CAM method.

1. The advantages and disadvantages of the methods mentioned in the references were not analyzed in the introduction section. The contributions of the proposed method were also not indicated. Moreover, why LSTM and KNN were combined was not explained.

2. The existing basic models were simply combined and applied to a new field in the paper. Moreover, the combination of KNN and LSTM models has been applied in other fields. Is there any further improvement?

3. The proposed model in the paper was only compared with the traditional models. It is better to compare it with the latest prediction models such as CNN-LSTM or Nbeats.

4. The sample size is around 300 in the experimental section, which is relatively small. Additionally, the training and testing set ratio is 9:1, which might introduce a potential bias or overfitting issue in the results. It would be beneficial to consider a larger sample size and evaluate the model's performance with a cross-validation approach to obtain more robust and reliable results.  

5. For the KNN algorithm, the selection of the K value is crucial. However, it was not described how to choose the appropriate K value to ensure the optimal performance of the model in the paper.

6. A large of existing basic theories were restated. The applications of the combined model mentioned in the paper should be discussed.

Moderate editing of English language required

Author Response

Tahnk you.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

the document is still meaningless and lacks both visual and application order, it should be reorganized in its entirety. Give an order in each structure of the applications as well as the results. It does not look like an orderly work. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

No more comments.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf