Using multivariate statistical analysis and Python programming for geochemical characterization at the Paraná Sedimentary Basin, Brazil

Characterizing the inherent heterogeneity of carbonate-rich shale sequences, such as the Irati Formation at the Paraná Sedimentary (PSB) in Brazil, is a complex challenge in petroleum geoscience. To address such task, a reproducible computational workflow has been developed integrating unsupervised multivariate statistical techniques - Principal Component Analysis (PCA) and Hierarchical Cluster Analysis (HCA) - for the objective interpretation of high-dimensional geochemical data. The workflow was systematically executed and, crucially, validated through implementation in two independent software environments (IBM SPSS and Python). Applied to a geochemical dataset from the Monte Olimpo mine in the Paraná Sedimentary Basin, the PCA reduced data dimensionality, extracting four components that explain 72.1% of the variance and delineate primary geochemical controls: a terrigenous-carbonate mixing axis (PC1) and a distinct radioactive association (PC2). HCA objectively classified elements and samples into coherent clusters, corroborating the PCA associations. The two platform validations confirmed the workflow’s robustness, with Python replication (using casewise deletion) yielding nearly identical results to the primary SPSS analysis. This integrated approach successfully identified two principal calcareous chemofacies with distinct signatures: Zone B is uniquely characterized by elevated radioactivity and SrO content compared to Zone A. Furthermore, sample clustering revealed geochemical mixing that cross-cut lithological boundaries, suggesting significant diagenetic overprinting. This study demonstrates that a clearly defined, reproducible computational workflow provides a powerful, transparent framework for decoding complex geochemical patterns, offering a reliable template for reservoir characterization that prioritizes methodological clarity and result verification.