Data for: The application of machine learning methods to aggregate geochemistry predicts quarry source location: a case study from the Irish aggregate industry.

Published: 27 April 2020| Version 1 | DOI: 10.17632/zzc668bbxz.1
Tadhg Dornan, Robbie Goodhue, Eva Stueeken


Attempting to classify the quarry sources which provided reactive rock aggregate, composed of Carboniferous aged pyritic mudrocks and limestones, to over 12, 500 homes across Ireland has not yet been possible using geochemical data. Using this dataset, a solution to this problem is found by applying machine learning models, such as logistic regression and random forest, to a geochemical dataset of scanning electron microscope energy-dispersive X-ray spectroscopy (SEM-EDS) and Laser ablation-quadrupole-inductively coupled plasma mass spectrometry (LA-Q-ICPMS) of pyrite, and Isotope ratio mass spectrometry (IRMS) of bulk rock aggregate, to predict quarry source location.