Application of Machine Learning Ensemble Methods to ASTRI Mini-Array Cherenkov Event Reconstruction
, Giancarlo Cusumano 1,3, Antonino La Barbera 1,3, Valentina La Parola 1,3 and Saverio Lombardi 3,4,5Round 1
Reviewer 1 Report
The article "Application of Machine Learning Ensemble methods to ASTRI Mini-Array Cherenkov event reconstruction" presents a detailed case study of gamma/hadron separation and energy reconstruction. They use a number of supervised Machine Learning methods for energy estimation.
The paper is interesting and after some corrections it can be of good quality.
The subsection 2.3 has strange structure. I suggest to rethink and rebuild the structure.
The authors mention comparing "a number of supervised Machine Learning methods, including the Random Forest method, Extra Trees method, Extreme Gradient Boosting (XGB)", however, in 2.3 only Random Forest is described. Why?
Figure 3 has some character sizes which can be considered too large. Maybe the whole figure could be smaller.
There are big differences in text character size between figures.
English language style, grammar, and readibility is fine, however, there are few minor grammatical problems/typos.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Dear Editor,
The article talks about the "Imaging Atmospheric Cherenkov technique". Despite being a subject that I don't have much expertise in, I will try to evaluate the article from the perspective of the machine learning techniques adopted, which I am more familiar with.
Overview: The article is a little long but well-written (I added minor corrections below). Some things could be moved to the supplementary material section.
The figures are well-designed and add value to the manuscript. Also, I think readers would benefit if the authors added a "conclusion" section.
Line 181 - why only include details about the random forest in the methods section? What about the other machine learning methods used?
I noticed that sections 3 and 4 seem part of the methods section (2). Why did you separate them?
How was cross-validation implemented?
Line 808 - Why can't the data be made available?
Minors:
Line 28 – hundreds => hundred
Line 28 – include a comma before “we can…”
Line 29 – include commas between “for example”
Line 29 – include a comma after “however”
Line 29 – include a comma before “which”
Line 47 – remove “the” before “scientific analysis”
Line 49 – did you mean “extended” instead of “exended”?
Line 55 – include a comma before “either”
Line 59 – include a comma after “Usually”
Line 61 – include a comma after “On the other hand”
Line 82 – “achieve an angular resolution”
Line 100 – “removing signals that ARE likely not related to”
Line 142 – “to the telescope’s positions on”
Line 144 – “The simulation consists of two subsets,”
Line 149 – Remove “on”
Line 151 – include a comma after “To this aim”
Line 153 – include a comma after “the second step”
Line 160 – “interest”
Line 162 – include a comma after “For each event”
Line 164 – “These also include”
Line 168 – “In the output,”
Line 286 and 288 – “tree-based”
Line 485 – “best-performing”
Line 532 – “is performed by applying”
Line 550 – include a comma after “benchmark”
Line 685 – remove the comma after “comparison”
Line 695 – “than ten and at least five times”
Line 725 – “of an energy bias”
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors addressed most of the issues presented in the review.
There is one important easy correction which can be done to achieve better quality.
The figures show very big differences in text size. Some figures have small label text size, other are very large. Please correct this to give a style for the paper.
Author Response
Thank you for your feedback on the review report. We have taken your comments into consideration and made the following adjustments:
-
Figure 1: We have enlarged the label text size to ensure better readability and consistency throughout the paper.
-
Figure 8: We have also increased the size of Figure 8.
-
Figure 7: Due to the high density of information, we have carefully evaluated the options. Unfortunately, modifying the figure to accommodate all the information (one plot per line) would result in excessive page length (two more pages).
We have made sure that the remaining figures in the paper are perfectly readable and informative. We appreciate your attention to detail and your concerns about the style of the paper. By making these adjustments, we believe we have addressed the issues raised while ensuring the figures are visually appealing, informative, and consistent with the overall style of the paper.
Author Response File: Author Response.pdf
