Ensemble ML framework predicts groundwater heavy metal pollution
Densu Basin groundwater faces heavy metal contamination that standard methods fail to model. The challenge: the Heavy Metal Pollution Index is skewed and entangles correlated contaminants, skewing predictions.
Researchers stacked six learners—SVM, k-NN, CART, Elastic Net, kernel ridge regression, and others—with three response transformations (raw, log, Gaussian copula) and nested cross-validation to capture spatial heterogeneity and statistical complexity.