Official repository: github: pyenfra
pyenfra is a Python library for fractal analysis, modeling and chaos estimation in time-series with the emphasis on environmental datasets. Package contains a suite of fractal and chaos metrics:
- Hurst exponent for rescaled range,
- Detrended Fluctuation Analysis (DFA),
- Multifractality by Generalized Hurst Slope,
- Wavelet Transform Modulus Maxima (with varying bands and modulus methods),
- Sample Entropy,
- Recurrence Quantification Analysis (RQA),
- Lyapunov exponents.
pyenfra can be installed using pip or locally by downloading package copy.
pip install:
pip install pyenfralocal:
use repositiory to obtain package copy.
Below are a couple examples of package usage.
Please refer to examples.py for extended, detailed examples and computation workflows.
import numpy as np
import matplotlib.pyplot as plt
import pyenfra
# Generate White noise sample data
ts_white = np.random.RandomState(0).randn(2000)
# Example: Compute Hurst exponent
h_value = pyenfra.functions.hurst(ts_white, num=30, min_n=10, min_segments=10)
# Example: Interpret Hurst
print(pyenfra.interpreters.interpret_hurst(ts_white, use_confidence_interval=False))
# Example: Plot Hurst climacogram for AR(1)
ax_hurst = pyenfra.plotting.plot_hurst(ts_white, num=30, min_n=10, min_segments=10,
figsize=(5,4), scatter_kwargs={'color':'C0'}, line_kwargs={'color':'C1'})
ax_hurst.figure.suptitle("Climacogram: AR(1) Persistent Process")
plt.show()
# Example: Compute Lyapunov Exponent
lyap_val, divergence, times = pyenfra.functions.lyapunov(ts_white, dim=3, tau=1, fs=1.0, max_iter=200, theiler=1)
print(f"Estimated Lyapunov exponent (logistic r=3.99): {lyap_val:.4f}")
print("Interpretation:", pyenfra.interpreters.interpret_lyapunov(lyap_val))Future works on the package include:
- HOST model integration.
- On demand functions.
Pull requests are welcome.
For major changes, please open an issue first to discuss implementation or changes.
This work was supported by NOAA grant NA19NOS4730207. Funding agency had no impact on work structure or findings.
This package is available under MIT license.