https://github.com/inoueakimitsu/cyclicmodel

Statistical causal discovery based on cyclic model
https://github.com/inoueakimitsu/cyclicmodel

causal-inference data-science probabilistic-programming pymc3 statistics

Last synced: about 2 months ago
JSON representation

Statistical causal discovery based on cyclic model

• Host: GitHub
• URL: https://github.com/inoueakimitsu/cyclicmodel
• Owner: inoueakimitsu
• License: mit
• Created: 2018-06-14T00:06:19.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-06-14T13:27:51.000Z (over 6 years ago)
• Last Synced: 2024-10-12T04:46:31.017Z (3 months ago)
• Topics: causal-inference, data-science, probabilistic-programming, pymc3, statistics
• Language: Python
• Size: 14.6 KB
• Stars: 2
• Watchers: 1
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# cyclicmodel

[![CircleCI](https://circleci.com/gh/inoueakimitsu/cyclicmodel/tree/master.svg?style=svg)](https://circleci.com/gh/inoueakimitsu/cyclicmodel/tree/master)

Statistical causal discovery based on cyclic model.  

This project is under development.

## Summary

Python package that performs statistical causal discovery

under the following condition:

1. there are unobserved common factors

2. two-way causal relationship exists

`cyclicmodel` has been developed based on

[`bmlingam`][4670f282], which implemented bayesian mixed LiNGAM.

  [4670f282]: https://github.com/taku-y/bmlingam "bmlingam"

## Example

```Python

import numpy as np

import pymc3 as pm

import cyclicmodel as cym

# Generate synthetic data,

# which assumes causal relation from x1 to x2

n = 200

x1 = np.random.randn(n)

x2 = x1 + np.random.uniform(low=-0.5, high=0.5, size=n)

xs = np.vstack([x1, x2]).T

# Model settings

hyper_params = cym.define_model.CyclicModelParams(

    dist_std_noise='log_normal',

    df_indvdl=8.0,

    dist_l_cov_21='uniform, -0.9, 0.9',

    dist_scale_indvdl='uniform, 0.1, 1.0',

    dist_beta_noise='uniform, 0.5, 6.0')

# Generate PyMC3 model

model = cym.define_model.get_pm3_model(xs, hyper_params, verbose=10)

# Run variational inference with PyMC3

with model:

  fit = pm.FullRankADVI().fit(n=100000)

  trace = fit.sample(1000, include_transformed=True)

# Check the posterior mean of the coefficients

print(np.mean(trace['b_21']))  # from x1 to x2

print(np.mean(trace['b_12']))  # from x2 to x1

```

## Installation

```bash

pip install cyclicmodel

```

## References

-  [LiNGAM - Discovery of non-gaussian linear causal models](https://sites.google.com/site/sshimizu06/lingam)

- [Shimizu, S., & Bollen, K. (2014). Bayesian estimation of causal direction in acyclic structural equation models with individual-specific confounder variables and non-Gaussian distributions. Journal of Machine Learning Research, 15(1), 2629-2652.](http://jmlr.org/papers/volume15/shimizu14a/shimizu14a.pdf)