This package implements gradient boosted classification and regression trees. In addition to piecewise constant base learners, piecewise linear regression trees are also implemented.
Gradient boosting is a supervised learning technique applicable to nonlinear regression and classification problems.
pCRTree is split into two subpackages:
- dtree: Classification and regression trees.
- boosting: Gradient boosting methods.
Supports N-Dimensional regression and classification.
Gradient boosted piecewise linear regression trees:
Example in example/boost_lin_tree_1d.py
Gradient boosted traditional regressing trees:
Quantile regression:
Example in example/classify_dblgauss.py
Depends:
- numpy
- scipy
- numba
Optional:
- matplotlib
- pytest (for testing)
- sklearn (for example and test benchmarks)
For developers:
$python3 setup.py develop --user
Users:
$python3 setup.py install --user