Hypertuning-Based Ensemble Machine Learning Approach for Real-Time Water Quality Monitoring and Prediction
, Habibur Rahman Rifat 1, Md Ashraf Uddin 2,*, Md Manowarul Islam 1, Md. Zulfiker Mahmud 1, Md Kowsar Hossain Sakib 3 and Arun Roy 1Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsPlease see the attachment.
Comments for author File: 
Comments.pdf
Author Response
Authors reply to reviewer is attached in the PDF.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsAlthough not novel, the theme of the article is interesting, and the content has relevant data. However, there is an apparent confusion with “prediction” and “determination”. Having available water quality parameters from an IoT-based sensor network, it is unclear the need for prediction mechanisms. For example, why the data presented in Fig. 7 are predictions? Are those results predictions “of future trends”? This question must be clarified.
The structure of the work is confusing, and the methodology must be clarified. Most readers with knowledge of water quality monitoring don’t understand IoT and ML techniques. Most readers with knowledge of these technologies don’t understand water quality parameters. So, an equilibrated short introduction to these themes can be useful.
Minor comments:
Machine Learning (ML) abbreviation must be defined in line 31.
Please clarify the sentence “The water quality … features” in lines 48-49.
The statement “In addition, … using WQI” must be supported by a reference to the literature.
Please ensure that all abbreviations are defined in their first use, such as AWS, RF, XGB, …, and check the repetition of definitions, such as IoT (lines 34, 96, 194, …).
Please justify why 97% accuracy is inadequate (line 115).
Please use subscripts and superscripts to present molecular formulas correctly in lines 148 to 162.
I recommend not starting section 3 with Fig. 1, but with text. Furthermore, I recommend that Figures and Tables must follow the citation in the text, and not be presented before the text.
The version and release of all software and applications must be indicated (section 3).
The model and manufacturer of the water parameter sensors must be presented (section 3).
The six water quality classes must be explained (lines 238 – 240).
Table 2 must be referred to in the text.
Please explain Table 2 data, because a total of more than 34,000 points are indicated in the text.
The maps in Fig. 2 are unreadable. Moreover, the difference between “Nested” and “Line” wells must be presented. Furthermore, the source of the images must be indicated.
I recommend elaborating on the “scaler” (line 250).
It’s not clear that all nine models were presented (lines 253-254). Furthermore, the method for model selection must be explained.
The WAI index must be explained, and some references indicated if applied (lines 258-259).
The weights, or the method for its determination, must be presented (lines 278-284). The WQI formula (equation) must be numbered.
Section 4.1 must be part of section 3.
Please check Fig. 5 (b), because only a line is shown.
I recommend shortening and merging the discussion section with the results section.
Author Response
Authors reply to reviewer is attached in the PDF.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe revised manuscript has addressed many of the concerns raised in the initial review. And the authors have made significant improvements in clarity and depth of analysis, which has enhanced the overall quality of the paper.
