multivariate_inference

This repository contains packages as well as code notebooks to test and compare various methods for performing inference tests on correlation matrices.

Parameter to vary

1. True correlation values np.arange(0, 1, .2) - 6 combinations
2. Sample size [10, 20, 30, 50, 100]- 5 combinations
3. Observation size [10, 100, 1000] - 3 combinations
4. Sampling structure [random, blocking, banding]- 3 combinations
5. Inference test [iid permutation, mantel test (2d permutation), cluster robust standard errors, linear mixed models, jacknife, jacknife second-order] - 6 combinations
6. Sample noise ?

Development

Can easily edit repo settings with the following make commands:

• make clean - remove all docs, build files, tests, and python artifacts
• make clean-docs - remove all built docs
• make test - run all tests using tox environments
• make docs - remake documentation HTML files
• make dist - create a pypi compatible distribution
• make install - install this package locally (i.e. using pip install -e .)

Installation

pip install git+https://github.com/cosanlab/multivariate_inference

Documentation

---

Credit

The scaffold for this package was created with Cookiecutter with the following template.