This package has been deprecated in favor of hbw.
hbw provides everything hessband does, plus:
- Multivariate support: KDE and Nadaraya-Watson for multivariate data
- More kernels: 6 kernels (gaussian, epanechnikov, uniform, biweight, triweight, cosine) vs 2
- Numba acceleration: 16-49× faster than pure NumPy implementations
- Large data handling: Automatic subsampling via
max_nparameter - Active development: Regular updates and improvements
pip uninstall hessband
pip install hbw# Old (hessband)
from hessband import select_kde_bandwidth, select_nw_bandwidth
# New (hbw)
from hbw import kde_bandwidth, nw_bandwidth- hbw repository: https://github.com/finite-sample/hbw
- hbw on PyPI: https://pypi.org/project/hbw/
- hbw documentation: https://finite-sample.github.io/hbw/