Distinguishing ore deposit type and barren sedimentary pyrite using laser ablation-inductively coupled plasma-mass spectrometry trace element data and statistical analysis of large data sets.

Simultaneous comparison of multiple trace elements through statistical analysis has been carried out using data from LA-ICP-MS pyrite trace element analysis and Random Forests (an ensemble machine-learning supervised classifier) to distinguish barren sedimentary pyrite and five ore deposit categories: iron oxide copper-gold (IOCG), orogenic Au, porphyry Cu, sedimentary exhalative (SEDEX), and volcanic-hosted massive sulphide (VHMS) deposits. The preferred classifier utilises in situ Co, Ni, Cu, Zn, As, Mo, Ag, Sb, Te, Tl, and Pb measurements to train the Random Forests. Testing of the Random Forests classifier using additional data from the same deposits and sedimentary basins (test data set) yielded an overall accuracy of 91.4% while testing using data from deposits and sedimentary basins that did not have analyses in the training data set yielded an overall accuracy of 88.0%. The performance of the classifier was further improved by instituting criteria to remove uncertain or inconclusive classifications, increasing the classifier’s accuracy to 94.5% for the test data and 93.9% for the blind test data. It is concluded that the Random Forests classification models for pyrite trace element data can be used for predictive modelling in greenfield terrains by providing an accurate indication of ore deposit type. The classifier will also be useful for palaeoenvironmental studies as it can accurately identify pyrite of sedimentary origin and so screen prospective samples that are affected by a hydrothermal overprint.
Simultaneous comparison of multiple trace elements through statistical analysis has been carried out using data from LA-ICP-MS pyrite trace element analysis and Random Forests (an ensemble machine-learning supervised classifier) to distinguish barren sedimentary pyrite and five ore deposit categories: iron oxide copper-gold (IOCG), orogenic Au, porphyry Cu, sedimentary exhalative (SEDEX), and volcanic-hosted massive sulphide (VHMS) deposits. The preferred classifier utilises in situ Co, Ni, Cu, Zn, As, Mo, Ag, Sb, Te, Tl, and Pb measurements to train the Random Forests. Testing of the Random Forests classifier using additional data from the same deposits and sedimentary basins (test data set) yielded an overall accuracy of 91.4% while testing using data from deposits and sedimentary basins that did not have analyses in the training data set yielded an overall accuracy of 88.0%. The performance of the classifier was further improved by instituting criteria to remove uncertain or inconclusive classifications, increasing the classifier’s accuracy to 94.5% for the test data and 93.9% for the blind test data. It is concluded that the Random Forests classification models for pyrite trace element data can be used for predictive modelling in greenfield terrains by providing an accurate indication of ore deposit type. The classifier will also be useful for palaeoenvironmental studies as it can accurately identify pyrite of sedimentary origin and so screen prospective samples that are affected by a hydrothermal overprint.