Prediction of Daily Water Consumption in Residential Areas Based on Meteorologic Conditions—Applying Gradient Boosting Regression Tree Algorithm

Round 1

Reviewer 1 Report

Based on meteorology factors (air temperature, ground temperature, humidity, air pressure, wind speed, etc.) and historical water consumption data in residential areas, the study uses correlation analysis to determine the factors that have the greatest impact on daily water consumption in residential areas. Finally, the relationship between these factors and water consumption is quantified through GBRT (Gradient boosting regression tree).

From this point of view, this research is an interesting methodological contribution to overcome the limitations of traditional statistical models due to their inability to deal with large data sets and complex structures common in the era of big data.

However, although a city in northeastern China is taken as a example of case study, sometimes it is difficult to correlate the theoretical analysis of this model with its practical translation into reality. It would be useful to break down further what is the reality to be analyzed and how the model is applied, specifying its results and highlighting its potential and limitations.

It would be appropriate for the authors, throughout the text, to provide more evidence, both at the level of data and of their interpretation and results, of the link between model and reality.

Related to this, the discussion and, particularly the conclusions, should also be developed more in this sense. The conclusions should be more precise, paying more attention to the results obtained and how the model contributes to obtaining a better prediction of consumption levels.

As a last point, perhaps some kind of reference should also be included on how the urban model and how its planning can influence in water consumption.

Author Response

attached

Author Response File: Author Response.pdf

Reviewer 2 Report

General remarks

Align left all the figures and figures captions and increase their resolution, that all the figures become well visible. Especially figures 1, 5, 6, 7, 8, 9 and 10.

Refer to the template when citing papers. The number of the reference is between square brackets and not as superscript.

All variables must be explained and represented italic in equations and in sentences too.

 

Minor remarks

Line 100. Briefly describe the content of each following sections.

Lines 106-115. The paragraph is too long.

Figure 1 is not visible. To be separated into two rows, with the “Gradient…” window on a second row.

Line 125. “The city involved in this study are located...” – either cities are, or city is.

Line 206. Here you define the abbreviation "General Formula for Somatosensory Temperature"(RGST)” and at Line 321 “Robert G.'s Somatosensory Temperature (RGST)”. Be consistent on that.

Line 262. Define w. Is it a mistake in notation?

Line 296. The table can be arranged not to go out of the page window.

Line 306. Define “Relative Humidity Precipitation”. Is it a “,” missing?

Line 324. Figure 3. Why not to add notations “DWC, RGST, SST, …, SD” under the horizontal axis? Explain the different shapes inside the heat map.

Line 336. Figure 4. Correct twice “Maxinum” with “Maximum”.

Line 374. Figure 6. No need for a legend when only one variable is represented.

Line 396. Explain the difference between the two notations: GI and Gini.

Line 415. “Observing Figure 12b” – correct, there is no figure 12b.

Lines 471-474. Review the sentence “This study discusses the influence of meteorologic factors on water consumption in residential areas and the possibility of using meteorologic factors to predict water consumption in residential areas and daily water consumption of residential areas can be predicted with an accuracy of ±8%.”.

Excellent paper, congratulations!

Review the logic of the sentences and the spelling.

Author Response

attached

Author Response File: Author Response.pdf

Reviewer 3 Report

First of all I want to congratulate the authors for their efforts in this manuscript. The paper presents and evaluates the performance of a gradient boost-ing regression tree algorithm to predict daily water consumption in residential areas based on meteorologic conditions. The topic is aligned with the scope of the journal and is relevant. In general terms, the paper is well structured and has good quality but there is a serious problem with the citation of references, which does not allow to follow properly some sections. Nevertheless, some aspects must be improved before accepting it. Following, I include a series of comments aimed at enhancing the quality of the paper:

1.       The abstract must start with a short description of the problem that the authors are investigating (maximum in two sentences). Then, the aim of the paper has to be established.

2.       Avoid using as a keyword, terms already used in the title. Deleted keywords included in the title and provided new keywords.

3.       There is a complete lack of references in the paper to contextualise it. Please provide references to justify the information included in the introduction. Check the citation format in the template and include the citations in the text.

4.       The introduction must mention the smart meters (and contextualise with a reference) since used datasets are collected with this type of device, and no mention of meters is made in the paper beyond line 138. Please add this reference (An integrated IoT architecture for smart metering).

5.       The aim of the paper must be presented in a new paragraph at the end of the introduction.

6.       The authors must provide a new version of Figure 1. General framework with better quality, it cannot be read properly. Consider enlarging the size of the figure and the letter size.

7.       Please check the letter format of Table 1 and follow the instructions and format of the template.

8.       In subsection 2.1.2 the authors can provide a general description of the used dataset, such as the number of data, period, and conducted data process between 2a and 2b. How was missing data obtained? Which technique for abnormal data detection is used?

9.       Considering that the paper is sent to Water (a journal not particularly focused on ML algorithms), it is strongly recommended to provide a schematic vision of gradient-boosted regression tree operation in subsection 2.3.

10.   In section 2, the authors have to add a section (or subsection) detailing the conducted steps in data processing and model creation. The authors have to detail the software used.

11.   In Figure 3. Weather factor correlation and significance heatmap I would like to recommend the authors to include box-whiskers diagrams for each parameter in the intersection where they include the correlation of each parameter with itself. Consider this as a mere suggestion to enhance the figure. It is not mandatory.

12.   The authors must extend the explanation of lines 333-335. Apparently, there is no correlation in Figure 3, and in Figure 4, it does not seem correlated. If the reason for including those parameters is based on literature, please add references.

13.   In Figure 7, the authors have to add the units in the axis names.

14.   In the discussion section, the authors have to compare their results with existing literature. Moreover, the limitations of the performed study must be highlighted and justified.

 

15.   Future work should be added in an independent paragraph at the end of the conclusions.

Author Response

attached

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

I want to congratulate the authors for their efforts in the review process. All the comments were correctly addressed.