Enhancing Geochemical Models with Compositional Data Analysis: A Systematic Review of Predictive Capabilities in Mineral Exploration and Environmental Assessment

Compositional data analysis (CoDA) has emerged as a critical methodology for enhancing the predictive capabilities of geochemical models in mineral exploration and environmental assessment applications. This systematic review examines the extent to which compositional data analysis techniques improve geochemical modeling performance in diverse geological settings and application domains. Through the analysis of 50 peer-reviewed studies, we identified 10 studies that met rigorous inclusion criteria, focusing on the quantitative evaluation of predictive modeling performance using compositional data analysis methods. The review reveals that log-ratio transformations, particularly centered log-ratio (CLR) and isometric log-ratio (ILR) transformations, consistently address the closure problem inherent in compositional geochemical data [1]. The integration of compositional data analysis with machine learning approaches, including random forests and principal component analysis, has demonstrated significant improvements in anomaly detection, geological class prediction, and mineralization identification. Quantitative results show classification accuracy improvements from 68.4% to 74.7% when maximum autocorrelation factor analysis is applied to compositionally transformed data compared to traditional principal component analysis (McKinley et al., 2018). The key findings indicate that compositional data analysis enhances predictive capabilities through improved data uniformity, more accurate anomaly identification, and better alignment with known geological processes. The methodology is particularly effective in diverse geological contexts, including stream sediment analysis, soil geochemistry, and regional mapping applications across multiple continents. However, this review identified limitations in reporting standards, with many studies lacking explicit accuracy metrics and baseline comparisons. This absence of standardized reporting significantly hinders the ability to compare findings across studies, assess the true impact of CoDA techniques, and draw robust and generalizable conclusions regarding their effectiveness. The inconsistent metrics make it challenging to synthesize quantitative results, potentially leading to an overestimation or underestimation of CoDA’s benefits in certain contexts. This systematic review also delves into a deeper analysis of these limitations, particularly the inconsistent reporting of quantitative performance metrics, and presents counterarguments to provide a more balanced perspective on the effectiveness of the compositional data analysis. This systematic review provides evidence that compositional data analysis significantly strengthens geochemical modeling capabilities, offering a robust framework for addressing the unique statistical challenges of geochemical datasets while improving practical outcomes in mineral exploration and environmental monitoring. However, this review also examines the limitations and potential counterarguments associated with the application of compositional data analysis in geochemical modeling, providing a more nuanced and balanced perspective. The author declares no conflicts of interest. The data supporting the conclusions of this article are available in the cited references. This study received no external funding.