Time Series Data Preparation for Failure Prediction in Smart Water Taps (SWT)
Round 1
Reviewer 1 Report
The submitted manuscript concerns the important issue of the time series data preparation for failure prediction in smart water taps. The paper presents a combination of the generative adversarial network (GAN) and the bidirectional gated recurrent unit (BiGRU) techniques for data preparation. The GAN aids in training the SWT data trend and distribution, enabling the imputed data to be closely similar to the historical dataset. On the other hand, the BiGRU was adopted to save computational time by combining the model's cell state and hidden state during data imputation. The result shows that this method can be applied in several time series systems to correct missing values in a dataset, thereby mitigating data noise that can lead to a biased failure prediction model. Furthermore, when evaluated using different sets of historical SWT data, the method proved reliable for missing data imputation and achieved better training time than the traditional data imputation method. Remarks: The relevance of the proposed methodology should be definitely discussed further. Line 79: the choice of reference should be supplemented with respect to the classification of missing data, as the source of missingness in a dataset is very important in data analysis because it affects the technique required to address the problem, missingness can be described as missing data values from a sample dataset (e.g. Ref. Modelling water distribution network failures and deterioration, 2017, IEEE International Conference on Industrial Engineering and Engineering Management 2017-December, 924-928. DOI 10.1109/IEEM.2017.8290027. The main achievements of this article should be presented and underlined in the conclusion. The conclusion should be specific. Please provide a deep analysis on conclusion. Also the pros and cons of the implemented method in comparison to others should be indicated. Try to underline the possible path for future studies. It is stated, that the manuscript is an Article, but the current version of your manuscript is not a scientific communication, it is rather like commentary, or can be classified like other form of publication. Authors do not explain well, where is the novelty of the distinguished method. Conclusion is not sufficiently described. It is more like summary of information, what can be read in previous chapters.
Author Response
The comments of reviewer 2 has been implemented in the revised manuscript
Reviewer 2 Report
Dear authors,
thanks for your contribution firstly.
The paper provides a study of a combination of the generative adversarial network (GAN) and the bidirectional gated recurrent unit (BiGRU) techniques for data preparation.
The abstract briefly summarizes the purpose of the paper and overall the article is well structured.
At the end of the introductory paragraph, I suggest that you better highlight the scientific novelty of your study. Even if the aim of the study is well explained it is important to highlight what the additional / new contribution to the research is.
In conclusion, I believe that the paper includes solid content, but some aspects need to be improved, improving them this manuscript can have its own value and impact.
I hope that these recommendations are helpful to the authors and wish good luck for the further reviewing process.
Author Response
The points raised have been implemented in the updated version of the article
Author Response File: 
Author Response.docx
Reviewer 3 Report
1. Abstract should be include more concise points on the time series data
2. Proof read. Line no 25, there is extra space in heading as well as . before "introduction"
3. Analysing time-series data and predicting future results of a time series dataset. "Improve English".
4. Across fieldssuch "Line no 27". Multiple errors on each lines.
5. Use the Journal template. Paper is not the required template
6. Introduction is vaguely presented around dataset. Needs to be detailed
7. Section 2 "Classification of Missing Data" does not make sense. Literature is not provided. Provide proper literature
8. Material is starting from the case study but it should start first from the dataset. Problem definition is talking mostly about data preparation
9. Cite figures in the text and explain them properly
10. 3.5.1 is actually like the introduction or literature providing information that GAN is proposed but what is the actual items proposed by the authors ? This section up to section 4 has too many items that does not make sense in the materials
11. Results and discussion should be separate and discussed in detail
12. Conclusion is different than the previous results
Author Response
All the reviews have been carefully implemented in the article
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Accept in the present form.
Author Response
There is no required correction from this reviewer
Reviewer 3 Report
Still many of previous comments not answered
1. Very few points have been improved on the introduction. Introduction is vaguely presented around dataset. Needs to be detailed
2. Abstract not improved
3. Proof read. Line no 25, there is extra space in heading as well as . before "introduction"
4. Across fieldssuch "Line no 27". Multiple errors on each lines.
5. Section 2 "Classification of Missing Data" does not make sense. Literature is not provided. Provide proper literature
6. Material is starting from the case study but it should start first from the dataset. Problem definition is talking mostly about data preparation
7. Cite figures in the text and explain them properly
8. Results and discussion should be separate and discussed in detail
Author Response
Response to the review queries
- Very few points have been improved on the introduction. Introduction is vaguely presented around dataset. Needs to be detailed
Response: The introduction has been significantly improved as can be seen with the tracked changes document compared to the final submission. We laid a foundation around the dataset because our work is based on the case study dataset. We also set a motion on the methods we used for the proposed BiGAN data imputation, Lines 54 to 57 mentions the impact of a poor dataset, while laying the ground for Line 63 which talks about selecting appropriate data imputation technique. Lines 78 to 87 talk about the contribution of our study.
- Abstract not improved
Response: The abstract has been improved to show that other improvements in the main body of the article
- Proof read. Line no 25, there is extra space in heading as well as . before "introduction"
Response: This has now been corrected as can be seen from the clean copy
- Across fieldssuch "Line no 27". Multiple errors on each lines.
Response: All errors have now been corrected from Lines 25 and beyond
- Section 2 "Classification of Missing Data" does not make sense. Literature is not provided. Provide proper literature
Response: This section has now been significantly improved and appropriate literature has been duly referenced
- Material starts from the case study but it should start first from the dataset. Problem definition is talking mostly about data preparation
Response: This has now been corrected
- Cite figures in the text and explain them properly
Response: Figures have been cited and described in the text properly
- Results and discussion should be separate and discussed in detail
Response: We kept results and discussion in one place so that we can effectively discuss our results in one place
Author Response File: Author Response.docx
