DOI
https://doi.org/10.1039/C9NJ05717J

Summary
The article focuses on introducing possible strategies to use machine learning (ML) algorithms for data mining in environmental chemistry. The authors describe several cases of ML acquisition. Firstly, the mixed toxicity analysis is considered to determine relationships between the pollutants and environmental or health risks. Secondly, the ML algorithms enable to process large amount of data, define complex relationships, and extract meaningful information even out of unsuccessful trials. Thus, ML can be employed in computational chemistry to rapidly screen the environmental functional material. Finally, the sample feature extraction out of visual or tabular data using ML is discussed. It is concluded that in order to employ ML at the reasonable level, researchers must produce data of good quality.

Relevance for Complex Systems Knowledge
The authors provide suggestions on how machine learning algorithms can be applied in the environmental chemistry. They state that complex systems usually are complicated and unordered. The main drawback in the research of environmental chemistry using conventional statistics is limited data to apply in analysis. On the contrary to conventional statistics, adding one more feature in the model does not require a significant amount of additional data. The phenomena of complex repetitive features results in the effectiveness of machine learning application. Although machine learning is useful in many areas in environmental chemistry, most of them are data driven, so it is still important to conduct real experiments and collect valuable data. The machine learning model can be applied to predict complex effects in mixture toxicity analysis, to analyze and screen materials which have nonlinear tendencies, such as nanocomposites, to perform quantitative and qualitative analysis, to extract features in the provided images.

Point of Strength
The strength of the article is a provided relationship between the machine learning methods and chemistry applications, such as analyzing mixture toxicity, environmental materials and feature extraction.