Dataset Information

Guo, Jie; ORCID: 0009-0007-4447-6296

This data provides a curated compilation of 650 experimental clinopyroxene samples published between 2021 and 2025, designed as a benchmark for evaluating machine learning-based clinopyroxene thermobarometers. The data are organized into three key tables: Table S1 contains the primary experimental data, including the major element compositions (12 oxides) of all 650 clinopyroxene samples, with pressure (P) and temperature (T) conditions. A subset of 214 samples also includes the corresponding co-existing melt compositions (13 oxides). Table S2 provides calculated Mahalanobis distances for each sample, offering a quantitative metric to assess whether a sample’s composition falls within or outside the training domains of existing machine learning-based clinopyroxene thermobarometers, which is critical for evaluating model extrapolation reliability. Table S3 presents a comparative analysis of pressure and temperature predictions generated by applying recent machine learning-based thermobarometers and a traditional thermobarometer pair to the compiled dataset, facilitating direct performance evaluation. This structured data package is essential for rigorously testing the accuracy and applicability limits of published thermobarometers and serves as a foundational resource for developing and calibrating new thermobarometer.

Guo, J., 2026. A compiled experimental dataset of clinopyroxene-liquid pairs (2021-2025) for evaluating machine learning-based clinopyroxene thermobarometers, Version 1.0. Interdisciplinary Earth Data Alliance (IEDA). https://doi.org/10.60520/IEDA/114342. Accessed 2026-04-09.

2026-02-19

Creative Commons Attribution 4.0 International [CC-BY-4.0]

Coverage Scope: Other

clinopyroxene, thermobarometer, machine Learning, evaluation, mahalanobis distance

2026-02-20

Collection