An unsupervised machine learning approach to iron meteorite classification

Abstract Iron meteorites are usually classified manually using bivariate elemental plots. This study extends iron meteorite classification into multi‐element space. We present a computational method for the classification of iron meteorites by applying unsupervised machine learning using density‐based cluster analysis. It provides formatted iron meteorite data extracted manually from 61 papers, as well as the software created for the application of cluster analysis to iron meteorites. The method can be used to speed up and standardize iron meteorite classification, reduce bias, increase transparency, and check reproducibility of existing classifications. It is fully scalable and can use any number and combination of elements. It allows for the classification of new iron meteorites, checks the validity of existing classifications, and can identify ungrouped iron meteorites that may be related to existing groups. The model has been applied to ungrouped iron meteorites and 29 are suggested for reclassification based on our multi‐element results.