Editors’ Highlights are summaries of recent papers by AGU’s journal editors.
Source: Geochemistry, Geophysics, Geosystems
Machine learning (ML) is becoming a common means of dealing with large, complex, multidimensional datasets in geochemistry. Tasks such as model regression, classification, clustering, and dimensionality reduction can all lead to insights into the structure of geochemical data that are difficult to obtain by traditional methods. However, there is a major barrier to entry for geochemists without training in data science or the ability to write code in Python.
ZhangZhou and He et al. [2024] introduce Geochemistry π, a new, easy-to-use tool that promises to make machine learning approaches to analysis of geochemical data accessible to those without coding skills. Geochemistry π seeks to overcome the barrier to entry by offering a step-by-step menu that guides the user through the choice and design of a machine learning study, with intelligent suggestion of default and automated parameter selection (but also control of parameter selection for more advanced users). The paper demonstrates that Geochemistry π successfully recovers a number of published ML results from recent geochemistry papers and explains how to use the tool.
Citation: ZhangZhou, J., He, C., Sun, J., Zhao, J., Lyu, Y., Wang, S., et al. (2024). Geochemistry π: Automated machine learning Python framework for tabular data. Geochemistry, Geophysics, Geosystems, 25, e2023GC011324. https://doi.org/10.1029/2023GC011324
—Paul Asimow, Editor, G-Cubed