A Comparative Study of Machine Learning Models with Hyperparameter Optimization Algorithm for Mapping Mineral Prospectivity

Selecting internal hyperparameters, which can be set by the automatic search algorithm, is important to improve generalization performance of machine learning models. In this study, geological, remote sensing and geochemical data Lalingzaohuo area in Qinghai province were researched. A multi-source metallogenic information spatial was constructed calculating Youden index for selecting potential evidence layers. The model mapping mineral prospectivity study established combining two swarm intelligence optimization algorithms, namely bat algorithm (BA) firefly (FA), with different receiver operating characteristic (ROC) prediction-area (P-A) curves used evaluation showed that algorithms had an obvious effect. BA FA differentiated improving multilayer perceptron (MLP), AdaBoost one-class support vector (OCSVM) models; thus, there no consistently superior other. However, accuracy models significantly enhanced after optimizing hyperparameters. under curve (AUC) values ROC optimized all higher than 0.8, indicating hyperparameter calculation effective. terms individual improvement, FA-AdaBoost improved most significantly, AUC value increasing from 0.8173 0.9597 prediction/area (P/A) 3.156 10.765, where targets predicted occupied 8.63% contained 92.86% known deposits. are consistent geological characteristics, combined efficient method mapping.