IsoScore 1.0

IsoScore measures uniformity of spatial utilization of point clouds

IsoScore

This contains the Python3 implementation of IsoScore, which was originally introduced in the 2021 paper by William Rudman, Nate Gillman, Taylor Rayne, and Carsten Eickhoff. IsoScore is a tool which measures how uniformly a point cloud utilizes the Euclidian space that it sits inside of. See the original paper for more information.

How to use

The only dependencies are numpy and sklearn.

import numpy as np
from IsoScore import IsoScore

# Computing the IsoScore for a fuzzy ball in R^3
point_cloud_isotropic = np.random.normal(size=(3,100))
the_score = IsoScore.IsoScore(point_cloud_isotropic)
print(f"The IsoScore for 100 points sampled from this Gaussian ball in R^3 is {the_score},")

# Computing the IsoScore for points sampled from the line t \mapsto (t, 2t, 3t) in R^3
random_array = np.random.normal(size=100)
point_cloud_anisotropic = np.array([random_array, 2*random_array, 3*random_array])
the_score = IsoScore.IsoScore(point_cloud_anisotropic)
print(f"and the IsoScore for 100 points sampled from this line in R^3 is {the_score}.")

License

This project is licensed under the MIT License.