Usage

Installation

Although the code is still highly experimental and in very active development, a release version is hosted in PyPI and can be installed using pip. The current state of the toolbox can be installed directly from the repository using git.

$ pip install unite_toolbox

Estimation

A typical example would involve the calculation of the entropy \(H(X)\) of a dataset. This can be accomplished through the k-NN estimator:

from scipy import stats
from unite_toolbox import knn_estimators

# Generate data
dist = stats.norm(loc=0.0, scale=0.6577)
samples = dist.rvs(size=(10_000, 1), random_state=42)

# Estimate entropy and print results
est_h = knn_estimators.calc_knn_entropy(samples)
print(f"Est. H = {est_h:.3f} nats")
print(f"True H = {dist.entropy():.3f} nats")

Est. H = 1.002 nats
True H = 1.000 nats

Success!