Predicting Power Consumption Using Deep Learning with Stationary Wavelet

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper need to be restructured and much improved. These are some of the points to be changed:

- Introductory part of the paper is quite long, but at the end of it there should be a list of the main contributions because it gives a better insight in what comes next

- x-axis of the Fig.2 is not clear. Does it contain data from random time instants in different years?

- In the table 1 we can see performance of the network. Values of RMSE, MAE and MAPE are given in what dimension? - it must be given in the table. Since MAPE is percentage error, the values given in the table are not acceptable!

- Fig. 3 has two figures. Second figure is zoomed part of the first one. It should be noted on the first figure which part is zoomed and given in the second figure. The same is for Fig. 4. and the others.

Tables from 8 to 18 contain very bad results with very large value of MAPE. It is sometimes good to give some bad results and to show how they are improved, but in this paper too long part of it is dedicated to bad results - it is not acceptable.

 

 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors,

Your paper is good, the analysis is proper, your methodology is full statistical.  As your reviewer, I have a few comments and I noticed some errors in the form of entries, especially percentages. Here are my comments/notes:

Note 1:Abstract: The line 12-13: he effectiveness of the SWT approach was evaluated using power consumption data at three different time intervals (1 min, 15 min, and 60 min). I didn't notice in the text that you defined the concept of effectiveness. If you use effectiveness, is it the relation of effects (income) to inputs (cost)? It would be better if you replaced this concept of efficiency with others. Moreover, were there only three intervals, because in the summary (conclusions) you do not write about these three intervals.

Note 2: Introduction: In my opinion, it is too long, please separate the text from the introduction that concerns the methodology and place it there. Moreover, in the Introduction, pose research questions RQs (RQ1, RQ2 , RQ3 (about3), the answers to RQ will be visible in the conclusions and results.

Note 3: Introduction again: is it your aim Line 134-135): This research delves into not only the overall efficacy of wavelet-based denoising but also underscores the essentiality of determining the most suitable decomposition level for this particular dataset.

If so, write that this is the goal of your analysis, and then put research questions under the goal: QR and, of course, define effectiveness, if you intended to measure it.

 

Note 4: Percentages, see, half of the text you write “percent” and half “%”, and here's another version “1.25% percent” (line 309).

 Note 5: Fig. Figure 3 and others. I would use "a" and "b" and write what is "a" and what is "b".

Note 6: References: In my opinion, they are too short, maybe because you do not have a literature review section, if other reviewers think that such a section is needed or a discussion section before the conclusion, they will complete the paper, add a section either: Background of analysis or Discussion. There are a lot of paper about your topic (examples of topics: Accurate Prediction of Power Consumption in Sensor Networks, Comparison of Machine Learning Algorithms for the Power Consumption Prediction, Assessment of Material Durability of Steam Pipelines Based on Statistical Analysis of Strength Properties etc.

the scope of analyzes is very wide, you can refer to them.

Best wishes

 Reviewer 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

This paper is quite improved, all the reviewer's comments are considered.