River Proximity Data as A Predictor for Ophiolite Classification: A Machine Learning Approach with OSM Data
Filip Arnaut · 2024
Summary The application of data-driven modeling and machine learning has attracted attention in the fields of geosciences and spatial modeling. Our previous research utilized the Random Forest (RF) and K-Nearest Neighbors (KNN) algorithms to forecast the presence of ophiolites in the East Vardar Ophiolite Zone (EVZ) in North Macedonia. This brief communication aims to assess the XGBoost model utilizing a random search for hyperparameter optimization and to incorporate a novel feature labeled “distance to river.” The application of the novel model and the new feature increased the ophiolite class F1-score by 0.14.