How do non-deposit sites influence the performance of machine learning-based gold prospectivity mapping? A study case in the Pitangui Greenstone Belt, Brazil
Brener Otávio Luiz Ribeiro , Danilo Barbuena , Gustavo Henrique Coelho de Melo , João Gabriel Motta , Eduardo Duarte Marques , Marcelo de Souza Marinho
{"title":"How do non-deposit sites influence the performance of machine learning-based gold prospectivity mapping? A study case in the Pitangui Greenstone Belt, Brazil","authors":"Brener Otávio Luiz Ribeiro , Danilo Barbuena , Gustavo Henrique Coelho de Melo , João Gabriel Motta , Eduardo Duarte Marques , Marcelo de Souza Marinho","doi":"10.1016/j.gexplo.2024.107543","DOIUrl":null,"url":null,"abstract":"<div><p>One of the greatest challenges in mineral prospectivity mapping (MPM) research nowadays is to find a solid methodology that ensures the reliability of the prospectivity model during the learning and prediction procedures. Multiple uncertainties such as the location of non-deposit sites or the type of machine learning algorithm (MLA) can bias the MPM. To investigate these effects, we used multiple training datasets with different non-deposits locations, randomly created, and MLAs such as Artificial Neural Network (ANN), Random Forests (RF) and Support Vector Machine (SVM), to model orogenic-Au prospectivity in the Pitangui Greenstone Belt (PGB, Brazil). Regarding the implications in the methodology for MPM, there are great differences between the models' performances in mapping prospective zones when there is a slightly change in the location of negative samples. These changes can be observed by using the Shapley additive explanation metrics (SHAP values), which can help mitigate such effects by choosing an optimal model among all randomly created datasets. The SHAP values of non-deposit sites also showed that ANN and SVM present overfitting problems despite the use of balanced data. RF on the other hand outperformed in all ten datasets and showed great recognition and adjustment to the negative samples. The results presented in this research are also promising to the prospective studies in the PGB, as it shows a map capable to correctly predict 97 % of the known deposits and occurrences in 3 % of the total area and points the new frontiers for gold exploration in the PGB.</p></div>","PeriodicalId":16336,"journal":{"name":"Journal of Geochemical Exploration","volume":null,"pages":null},"PeriodicalIF":3.4000,"publicationDate":"2024-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Geochemical Exploration","FirstCategoryId":"89","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0375674224001596","RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"GEOCHEMISTRY & GEOPHYSICS","Score":null,"Total":0}
引用次数: 0
Abstract
One of the greatest challenges in mineral prospectivity mapping (MPM) research nowadays is to find a solid methodology that ensures the reliability of the prospectivity model during the learning and prediction procedures. Multiple uncertainties such as the location of non-deposit sites or the type of machine learning algorithm (MLA) can bias the MPM. To investigate these effects, we used multiple training datasets with different non-deposits locations, randomly created, and MLAs such as Artificial Neural Network (ANN), Random Forests (RF) and Support Vector Machine (SVM), to model orogenic-Au prospectivity in the Pitangui Greenstone Belt (PGB, Brazil). Regarding the implications in the methodology for MPM, there are great differences between the models' performances in mapping prospective zones when there is a slightly change in the location of negative samples. These changes can be observed by using the Shapley additive explanation metrics (SHAP values), which can help mitigate such effects by choosing an optimal model among all randomly created datasets. The SHAP values of non-deposit sites also showed that ANN and SVM present overfitting problems despite the use of balanced data. RF on the other hand outperformed in all ten datasets and showed great recognition and adjustment to the negative samples. The results presented in this research are also promising to the prospective studies in the PGB, as it shows a map capable to correctly predict 97 % of the known deposits and occurrences in 3 % of the total area and points the new frontiers for gold exploration in the PGB.
期刊介绍:
Journal of Geochemical Exploration is mostly dedicated to publication of original studies in exploration and environmental geochemistry and related topics.
Contributions considered of prevalent interest for the journal include researches based on the application of innovative methods to:
define the genesis and the evolution of mineral deposits including transfer of elements in large-scale mineralized areas.
analyze complex systems at the boundaries between bio-geochemistry, metal transport and mineral accumulation.
evaluate effects of historical mining activities on the surface environment.
trace pollutant sources and define their fate and transport models in the near-surface and surface environments involving solid, fluid and aerial matrices.
assess and quantify natural and technogenic radioactivity in the environment.
determine geochemical anomalies and set baseline reference values using compositional data analysis, multivariate statistics and geo-spatial analysis.
assess the impacts of anthropogenic contamination on ecosystems and human health at local and regional scale to prioritize and classify risks through deterministic and stochastic approaches.
Papers dedicated to the presentation of newly developed methods in analytical geochemistry to be applied in the field or in laboratory are also within the topics of interest for the journal.