Publication
Overview
The study compares distance-weighted nearest neighbors, boosted decision trees, random forests, artificial neural networks, and support vector machines for CPT-based soil classification and combines them through a heterogeneous majority-voting multiple model predictor.
Code lineage and reproducibility
MMP is the principal public companion repository for the multiple-model ensemble. The recovered source predates the final article and is preserved with explicit provenance notes, including the distinction between the paper's SMOTE description and the balancing behavior found in the recovered 2020 scripts.
Two additional repositories document the historical component-model implementations. Trees preserves/refactors the decision-tree and random-forest predecessor source, while ANN_SVM preserves the earlier ANN/SVM development source and deterministic wrappers. These should be read as historical code lineage, not as claims of byte-for-byte final-paper reproduction.