Assessing the Suitability of Boosting Machine-Learning Algorithms for Classifying Arsenic-Contaminated Waters: A Novel Model-Explainable Approach Using SHapley Additive exPlanations
Round 1
Reviewer 1 Report
This is very interesting paper on popular boosting algorithms , some newly introduced (NGB, CATB, 583 and ADAB) on the problem of arsenic modeling. Authors have used smaller , regional, dataset from available source ( Ghana) This could be a potential limitation , however , in my opinion think , beside the limitation that local research gives, this could be worthy model, if proven by larger research, or to be considered a pretext for future , widerscale, research. The study is also explaining model decisions in order to unravel the complex underlying non-linear relationship between influencing differnt input variables and this study provides a comprehensive evaluation of boosting algorithms and explainable AI . All of this cab be used inpredicition and general categorisation of arsenic in different water supply systems. This kind of research has potential, and is , at least , in this small scale, regional research, considered predictive .
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Valuable research has been done and interesting results have been obtained that can be considered in the use of machine learning and explainable methods for arsenic modelling. The purpose and hypothesis of the article are not well stated in the introduction, which needs to be pointed out precisely. In my opinion, this article is worthy of acceptance, however, there are some minor corrections listed below that need to be made.
- Although the article is relatively well written, it is necessary to do a general review and fix the grammatical and writing errors.
- The method of measuring arsenic with ICP-MS device and the preparation steps of treatment and control samples should be fully described.
- The discussion is well written and thus could better address the results obtained. The discussion needs to be improved.
- The conclusion section should be rewritten to more accurately address the findings of this research
Author Response
Please see the attachment.
Author Response File: Author Response.pdf