Skip to main content

A python library for building different types of copulas and using them for sampling.

Project description

“Copulas” An open source project from Data to AI Lab at MIT.

Copulas

Overview

A python library for building different types of copulas and using them for sampling.

Supported Copulas

Bivariate

  • Clayton
  • Frank
  • Gumbel

Accesible from copulas.bivariate.copulas.Copula

Multivariate

Accesible from copulas.multivariate.models.CopulaModel

Installation

Install with pip

The easiest way to install Copulas is using pip

pip install copulas

Install from sources

You can also clone the repository and install it from sources

git clone git@github.com:DAI-Lab/Copulas.git
cd Copulas
python setup.py install

Data Requirements

This package works under the assumption that the data is perfectly clean, that means that:

  • There are no missing values.
  • All values are numerical

Usage

In this library you can model univariate distributions and create copulas from a numeric dataset. For this example, we will use the iris dataset in the data folder.

Creating Univariate Distribution

First we will retrieve the data from the data folder and create a univariate distribution. For this example, we will create a normal distribution. First type the following commands on a python terminal.

>>> from copulas.univariate.gaussian import GaussianUnivariate
>>> import pandas as pd
>>> data = pd.read_csv('data/iris.data.csv')
>>> data.head()
   feature_01  feature_02  feature_03  feature_04
   0         5.1         3.5         1.4         0.2
   1         4.9         3.0         1.4         0.2
   2         4.7         3.2         1.3         0.2
   3         4.6         3.1         1.5         0.2
   4         5.0         3.6         1.4         0.2

Once we have the data, we can pass it into the GaussianUnivariate class.

>>> feature1 = data['feature_01']
>>> gu = GaussianUnivariate()
>>> gu.fit(feature1)
>>> print(gu)
Distribution Type: Gaussian
mean =  5.843333333333335
standard deviation =  0.8253012917851409
max =  7.9
min =  4.3

Once you fit the distribution, you can get the pdf or cdf of data points and you can sample from the distribution.

>>> gu.get_pdf(5)
0.28678585054723732
>>> gu.get_cdf(5)
0.15342617720079199
>>> gu.sample(1)
array([ 6.14745446])

Creating a Gaussian Copula

When you have a numeric data table, you can also create a copula and use it to sample from the multivariate distribution. In this example, we will use a Gaussian Copula.

>>> from copulas.multivariate.gaussian import GaussianMultivariate
>>> gc = GaussianMultivariate()
>>> gc.fit(data)
>>> print(gc)
feature_01
===============
Distribution Type: Gaussian
Variable name: feature_01
Mean: 5.843333333333334
Standard deviation: 0.8253012917851409
Max: 7.9
Min: 4.3

feature_02
===============
Distribution Type: Gaussian
Variable name: feature_02
Mean: 3.0540000000000003
Standard deviation: 0.4321465800705435
Max: 4.4
Min: 2.0

feature_03
===============
Distribution Type: Gaussian
Variable name: feature_03
Mean: 3.758666666666666
Standard deviation: 1.7585291834055212
Max: 6.9
Min: 1.0

feature_04
===============
Distribution Type: Gaussian
Variable name: feature_04
Mean: 1.1986666666666668
Standard deviation: 0.7606126185881716
Max: 2.5
Min: 0.1

Copula Distribution:
     feature_01  feature_02  feature_03  feature_04
0     -0.900681    1.032057   -1.341272   -1.312977
1     -1.143017   -0.124958   -1.341272   -1.312977
2     -1.385353    0.337848   -1.398138   -1.312977
3     -1.506521    0.106445   -1.284407   -1.312977
4     -1.021849    1.263460   -1.341272   -1.312977
5     -0.537178    1.957669   -1.170675   -1.050031
...

[150 rows x 4 columns]

Covariance matrix:
[[ 1.26935536  0.64987728  0.94166734 ... -0.57458312 -0.14548004
  -0.43589371]
 [ 0.64987728  0.33302068  0.4849735  ... -0.29401609 -0.06772633
  -0.21867228]
 [ 0.94166734  0.4849735   0.72674568 ... -0.42778472 -0.04608618
  -0.27836438]
 ...
 [-0.57458312 -0.29401609 -0.42778472 ...  0.2708685   0.0786054
   0.19208669]
 [-0.14548004 -0.06772633 -0.04608618 ...  0.0786054   0.17668562
   0.14455133]
 [-0.43589371 -0.21867228 -0.27836438 ...  0.19208669  0.14455133
   0.22229033]]

Means:
[-3.315866100213801e-16, -7.815970093361102e-16, 2.842170943040401e-16, -2.3684757858670006e-16]

Once you have fit the copula, you can sample from it.

gc.sample(5)
   feature_01  feature_02  feature_03  feature_04
0    5.529610    2.966947    3.162891    0.974260
1    5.708827    3.011078    3.407812    1.149803
2    4.623795    2.712284    1.283194    0.213796
3    5.952688    3.086259    4.088219    1.382523
4    5.360256    2.920929    2.844729    0.826919

History

0.1.1 - Minor Improvements

  • Different Copula types separated in subclasses
  • Extensive Unit Testing
  • More pythonic names in the public API.
  • Stop using third party elements that will be deprected soon.
  • Add methods to sample new data on bivariate copulas.
  • New KDE Univariate copula
  • Improved examples with additional demo data.

0.1.0 - First Release

  • First release on PyPI.

Project details


Release history Release notifications | RSS feed

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

copulas-0.1.1.tar.gz (28.6 kB view details)

Uploaded Source

Built Distribution

copulas-0.1.1-py2.py3-none-any.whl (20.6 kB view details)

Uploaded Python 2Python 3

File details

Details for the file copulas-0.1.1.tar.gz.

File metadata

  • Download URL: copulas-0.1.1.tar.gz
  • Upload date:
  • Size: 28.6 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/1.11.0 pkginfo/1.4.2 requests/2.19.1 setuptools/40.0.0 requests-toolbelt/0.8.0 tqdm/4.24.0 CPython/3.6.6

File hashes

Hashes for copulas-0.1.1.tar.gz
Algorithm Hash digest
SHA256 d25166d91b1f2c90c4e4edcc4533f7357798c712d0e8bec321d91eae1a295e79
MD5 139c5df5767586c4bc58f6d0a7bc60a8
BLAKE2b-256 9621af9707c9d175ed54a30d0194d10aeed1f9b37148596edbfb041ea11a8933

See more details on using hashes here.

File details

Details for the file copulas-0.1.1-py2.py3-none-any.whl.

File metadata

  • Download URL: copulas-0.1.1-py2.py3-none-any.whl
  • Upload date:
  • Size: 20.6 kB
  • Tags: Python 2, Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/1.11.0 pkginfo/1.4.2 requests/2.19.1 setuptools/40.0.0 requests-toolbelt/0.8.0 tqdm/4.24.0 CPython/3.6.6

File hashes

Hashes for copulas-0.1.1-py2.py3-none-any.whl
Algorithm Hash digest
SHA256 7c368486b7fff8cbd67ae7d896097dbab99268d2a0f32a246dd244a2c04f7df3
MD5 fbd1bb3305b0b70f59b3fe5990b7ad76
BLAKE2b-256 779de1717bd9935c51d46570372e347eaacf37b6a6c5a14d2485a62b6e3bcede

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page