Data

Tools

Indexes

Applications

Experiments

Open Source Science

Recent Releases of infomeasure

infomeasure - Statistical Testing, New Entropy Estimators and Documentation

This release introduces version 0.5.0 of infomeasure, featuring a comprehensive documentation overhaul, enhanced statistical testing capabilities, and 10+ new entropy estimators.

Breaking Changes

  • Statistical Testing API Overhaul: Replaced p_value() and t_score() methods with comprehensive statistical_test() method returning rich StatisticalTestResult objects
  • Configuration Updates: Renamed p_value_method to statistical_test_method and added statistical_test_n_tests parameter

Key Highlights

Enhanced Statistical Testing Framework

  • New StatisticalTestResult Class: Rich result objects with p-values, t-scores, confidence intervals, and metadata
  • Flexible Analysis: On-demand percentile calculation and statistical analysis
  • Improved API: Simplified interface with structured return types

New Entropy Estimators

Added comprehensive suite of discrete entropy estimators based on meta-analysis research:

  • Miller-Madow Suite: Bias-corrected estimators for H, MI, CMI, TE, CTE, KLD, and JSD
  • Bayesian Entropy: Multiple prior choices (Jeffrey, Laplace, Schurmann-Grassberger, Minimax)
  • Chao-Shen, Shrinkage, Grassberger Entropy: Specialized bias correction techniques
  • NSB & ANSB Entropy: Bayesian estimates for undersampled data
  • Zhang & Bonachela Entropy: Optimized for small datasets

Documentation Improvements

  • New Estimator Selection Guide: Interactive Q&A approach for choosing optimal estimators
  • New Statistical Tests Guide: Complete framework for hypothesis testing
  • Enhanced Existing Guides: Major expansions with performance visualizations and research-based recommendations

Installation

You can install or upgrade the package using: bash pip install infomeasure --upgrade

Documentation

For complete details and usage examples, visit the documentation.

New Guides: - Estimator Selection Guide - Interactive approach to choosing the right estimator - Statistical Tests Guide - Comprehensive statistical testing framework

Changelog: Full changelog

This release aims to provide a more comprehensive and research-backed toolkit for information-theoretic analysis with improved documentation and user experience.

All Commits: https://github.com/cbueth/infomeasure/compare/0.4.0...0.5.0

- Python
Published by cbueth 8 months ago

infomeasure - Cross-Entropy Support and Stability Improvements

This release introduces cross-entropy support, documentation enhancements, and stability improvements, marking the project as stable.

Key Highlights

1. Cross-Entropy Support

  • Added cross-entropy calculations for all approaches.
  • Restricted joint random variables (RVs) for cross-entropy to ensure clarity.

2. Documentation Updates

  • Added detailed documentation for cross-entropy, with examples and integration into the existing documentation structure.
  • Enhanced README.md with logos, badges, and improved formatting.

3. Stability and Code Quality

  • Addressed warnings in the code by treating them as errors in pytest.
  • Improved OS compatibility for tests to ensure broader support.

Changelog

For a complete list of changes, see the changelog.

Installation

You can install or upgrade the package using: bash pip install infomeasure --upgrade

Acknowledgments

For more details, visit the documentation.

Commits: https://github.com/cbueth/infomeasure/compare/0.3.3...0.4.0

- Python
Published by cbueth 10 months ago

infomeasure - First GitHub Release: Enhanced Information Measures and Optimizations

This is the first GitHub release of the infomeasure package, now versioned with Zenodo. The release includes essential information measures such as Entropy (H), Mutual Information (MI), Conditional Mutual Information (CMI), Transfer Entropy (TE), and more. Highlights include:

  • Support for local values, optimized for discrete and ordinal data.
  • New composite measures: Jensen-Shannon Divergence (JSD) and Kullback-Leibler Divergence (KLD).
  • Performance improvements for TE, CTE, MI, and CMI estimators with vectorized implementations.
  • Major API updates for compatibility with arbitrary random variables.
  • Enhanced tests and validation, ensuring robust functionality.

Changelog

To get to know more about infomeasure feel free to read the documentation, now hosted at Read the Docs.

The package supports Python 3.11-3.13, includes a CI/CD pipeline, and is licensed under AGPLv3+.

All Commits: https://github.com/cbueth/infomeasure/commits/0.3.3

- Python
Published by cbueth 10 months ago