Machine learning applied to rock geochemistry for predictive outcomes: The Neapolitan volcanic history case

Abstract

In this paper we explore the efficiency of various machine learning techniques to determine the volcano source,the eruptive formation and the eruption period of volcanic rocks when their chemical contents are known. Withthis aim, we assembled a data set of 9800 volcanic rocks from the open-access literature. The rocks belong toeruptive formations from Somma-Vesuvius, Campi Flegrei, Ischia and Procida volcanoes, in the Neapolitan regionof Italy.The data set includes content of majoroxidesand trace elements,aswell asSrand Nd isotope ratios,erup-tive periods, eruption formations and volcano source. Some discrete numerical variables are missing in certainsamples resulting in data exclusion and measurement inhomogeneity. Our results indicate that, despite such is-sues, some machine learning algorithms have a very high prediction ability, i.e., at >70%. The achieved resultsare interesting in order to facilitate the managing of new data for volcanological reconstruction andtephrostratigraphic studies