09:00 〜 10:30
[MGI29-P02] 深海性かんらん岩単斜輝石中の微量元素組成における組成データ解析
キーワード：単斜輝石、マントル、かんらん岩、統計解析
Data analysis using statistical and machine learning methods has recently been applied to geochemical data (Itano et al., 2020). We newly applied compositional data analysis (CoDA), including log-ratio transformation, principal component analysis, and k-means clustering on a large database of abyssal peridotite clinopyroxene composition and showed great potential for understanding compositional systematics (Nishio et al., 2022).
CoDA is a set of multivariate statistical analyses and is used for major element data (Aitchison, 1982). Compositional data consist of a matrix of nonnegative, relative values with a constant. In the Earth sciences, mineral chemical compositions are compositional data, and mineral trace element compositions are the parts of compositional data which is called subcomposition. Concentrations of various elements within a mineral are dependent on the concentrations of other elements in the mineral because the compositional data has a constant sum. Therefore, n-dimensional data are plotted on an n−1 dimensional space, as the concentration of one element is not an independent variable, and the potential for pseudo-correlations should be considered. CoDA considers and removes these constant-sum constraints on the following statistical analyses. We applied these methods to clinopyroxene trace element compositions, which are subcompositional data. In this presentation, we will introduce the analyses, performed by Nishio et al. (2022) and discuss the importance of log-ratio transformations in the statistical analysis of clinopyroxene trace elements.
References
Itano, K., Ueki, K., Iizuka, T., & Kuwatani, T. (2020). Geochemical discrimination of Monazite source rock based on machine learning techniques and multinomial logistic regression analysis. Geosciences, 10(2), 63. https://doi.org/10.3390/geosciences10020063
Nishio, I., Itano, K., Waterton, P., Tamura, A., Szilas, K., & Morishita, T. (2022). Compositional Data Analysis (CoDA) of Clinopyroxene From Abyssal Peridotites. Geochemistry, Geophysics, Geosystems, 23(8), e2022GC010472. https://doi.org/10.1029/2022GC010472
Aitchison, J. (1982). The statistical analysis of compositional data. Journal of the Royal Statistical Society: Series B, 44(2), 139– 160. https://doi.org/10.1111/J.2517-6161.1982.TB01195.X
CoDA is a set of multivariate statistical analyses and is used for major element data (Aitchison, 1982). Compositional data consist of a matrix of nonnegative, relative values with a constant. In the Earth sciences, mineral chemical compositions are compositional data, and mineral trace element compositions are the parts of compositional data which is called subcomposition. Concentrations of various elements within a mineral are dependent on the concentrations of other elements in the mineral because the compositional data has a constant sum. Therefore, n-dimensional data are plotted on an n−1 dimensional space, as the concentration of one element is not an independent variable, and the potential for pseudo-correlations should be considered. CoDA considers and removes these constant-sum constraints on the following statistical analyses. We applied these methods to clinopyroxene trace element compositions, which are subcompositional data. In this presentation, we will introduce the analyses, performed by Nishio et al. (2022) and discuss the importance of log-ratio transformations in the statistical analysis of clinopyroxene trace elements.
References
Itano, K., Ueki, K., Iizuka, T., & Kuwatani, T. (2020). Geochemical discrimination of Monazite source rock based on machine learning techniques and multinomial logistic regression analysis. Geosciences, 10(2), 63. https://doi.org/10.3390/geosciences10020063
Nishio, I., Itano, K., Waterton, P., Tamura, A., Szilas, K., & Morishita, T. (2022). Compositional Data Analysis (CoDA) of Clinopyroxene From Abyssal Peridotites. Geochemistry, Geophysics, Geosystems, 23(8), e2022GC010472. https://doi.org/10.1029/2022GC010472
Aitchison, J. (1982). The statistical analysis of compositional data. Journal of the Royal Statistical Society: Series B, 44(2), 139– 160. https://doi.org/10.1111/J.2517-6161.1982.TB01195.X