pyanodi

Pyanodi is a Python package to perform the analysis of distance designed by Tan et al. (2014) to compare geostatistical simulation algorithms.

Disclaimer

Pyanodi only contains the cluster-based histograms of patterns but not the multiple-point histograms.

Installation

You can directly install pyanodi from GitHub using pip:

pip install git+https://github.com/grongier/pyanodi.git

Use

Basic use:

from pyanodi import ANODI
from sklearn.manifold import MDS

training_image = ... # nD array
realizations = ...   # (n_methods, n_realizations_per_method, nD) array

# Set the parameters
anodi = ANODI(pyramid=(1, 2, 3),
              random_state=42,
              n_jobs=4)

# Perform the analysis
anodi.fit_transform(training_image, realizations)

# Get the MDS representation of the distances between the images
pyramid_level = 0
mds = MDS(n_components=2, dissimilarity='precomputed', random_state=100)
mds_points = mds.fit_transform(anodi.distances_[..., pyramid_level])

# Get the rankings of the methods
anodi.score()

For a more complete example, see the Jupyter notebook methods_comparison.ipynb or the Binder link above.

Citation

If you use pyanodi in your research, please cite the original article:

Tan, X., Tahmasebi, P. & Caers, J. (2014). Comparing Training-Image Based Algorithms Using an Analysis of Distance. Mathematical Geosciences, 46(2), 149-169. doi:10.1007/s11004-013-9482-1

Here is the corresponding BibTex entry if you use LaTex:

@Article{Tan2014,
    author="Tan, Xiaojin
    and Tahmasebi, Pejman
    and Caers, Jef",
    title="Comparing Training-Image Based Algorithms Using an Analysis of Distance",
    journal="Mathematical Geosciences",
    year="2014",
    volume="46",
    number="2",
    pages="149--169",
    issn="1874-8953",
    doi="10.1007/s11004-013-9482-1",
}