SVM On Trees

Title: SVM on Trees This record provides the reproducible implementation (C++17 with Python/pybind11 bindings) for the working paper/preprint “SVM on Trees”. Version v1.1.0 updates the experimental evaluation by adding five baseline models, additional datasets, repeated stratified train/test splits, statistical tests, and data separability analysis. Repository (code): https://github.com/Tran-Thu-Le/P2507-SVM-on-tree.gitPreprint (HAL): https://hal.science/hal-05505952PDF (HAL v1): https://hal.science/hal-05505952v1/file/P25-SVM-on-tree.pdf Contents Core C++ implementation and Python bindings (pybind11) Benchmark comparing SVM-on-Tree with LinearSVC, RBF-SVM, Decision Tree, Random Forest, and Gradient Boosting Experiments on Synthetic, Wine, Breast Cancer, Digits, and MAGIC Gamma Telescope datasets Repeated stratified evaluation and Wilcoxon signed-rank tests Data separability and spine-suitability analysis Updated dependencies and reproduction instructions Reproducible environment via requirements.txt License: CC BY 4.0