USING DATA SCIENCE METHODS TO EXPLORE THE ORIGINS OF UNGROUPED CARBONACEOUS CHONDRITES

Ungrouped chondrites are a unique subset of carbonaceous chondrites that do not fit into any of the currently classified groups. While some of these chondrites are unique specimens, others exhibit certain relationships to each other and/or to established groups. The classification of meteorites, based on their chemical composition, texture, mineral compositions, and modal abundances, is successful for most samples. However, the ungrouped samples highlight the need for deeper exploration of the available data. Building a dataset that includes the chemical composition of all carbonaceous chondrites, with particular emphasis on comprehensive data for ungrouped chondrites, presents an excellent and challenging opportunity for testing existing classifications and potentially revealing valuable insights into the origin of ungrouped members. Multivariable analysis offers a more in-depth approach that is not limited to the number of variables or visualizing results in 2D-3D space for interpretation.

Ungrouped chondrite data entails several additional steps to ensure its usability. Data cleaning, combining different datasets, unit conversion, and data normalization are necessary to ensure that the data is in a usable format. Furthermore, the data analysis process requires creative thinking and inventive approaches to data processing, given the incomplete data that makes applying standard techniques like cluster analysis for classification challenging. This challenge arises due to the significant number of missing data caused by various factors such as the format of reporting data, limitations of equipment and methods, or the nature of the material and low concentration.

Despite these challenges, data science methods offer a unique opportunity to explore the chemical composition of ungrouped chondrites. Machine learning algorithms and network analysis provide alternative methods for exploring relationships and similarities between ungrouped chondrites and other chondrites without the need to discard data due to limited variables. The application of data science in cosmochemistry shows promise for exploring extraterrestrial materials and advancing our understanding of the origin and history of the Solar System.