VineCopulas

VineCopulas is a Python package that is able to: * Fit both bivariate and vine copulas * Simulate from both bivariate and vine copulas * Allow for both discrete as well as continuous input data * Draw conditional samples for any variables of interest with the use of bivariate copulas and different vine structures

Installation

pip install vinecopulas

Getting Started

Get started by testing out the package functionality using the Abalone example data.

```python import pandas as pd import matplotlib.pyplot as plt import numpy as np from vinecopulas.marginals import * from vinecopulas.bivariate import * from vinecopulas.vinecopula import *

datapath = 'https://raw.githubusercontent.com/VU-IVM/vinecopula/develop/doc/sampledata.csv' df = pd.readcsv(datapath) df.head() ```

Transform the data to pseudo data and fit a survival gumbel copula between two variables. Use the fitted copula to generate random samples.

```python x = np.array(df)[:,:-1] u = pseudodata(x) # computes the pseudodata cop = 4 # copula 4 is the gumbel copula with 180 degree rotation par = fit(cop, u[:,:2]) # fit the variables in the first 2 columns n = len(u) # number of samples to generate ur = random(cop, par, n) # generate random samples

plot

plt.scatter(u[:,0],u[:,1], label = 'Data') plt.scatter(ur[:,0], ur[:,1], alpha = 0.5, label = 'Random') plt.xlabel('$u1$') plt.ylabel('$u2$') plt.legend() ```

Fit a vine copula between multiple variables in the data, considering all possible copulas available in the package.

python cops = list(range(1,16)) # fit vine copula according to these copulas M, P, C = fit_vinecop(u, cops, vine = 'R') # fit R-vine plotvine(M,variables = list(df.columns[:-1]), plottitle = 'R-Vine') # plot structure

For more examples, please have a look at the documentation.

Contribution Guidelines

Please look at our Contributing Guidelines if you are interested in contributing to this package.

Asking Questions and Reporting Issues

If you encounter any bugs or issues while using VineCopulas, please report them by opening an issue in the GitHub repository. Be sure to provide detailed information about the problem, such as steps to reproduce it, including operating system and Python version.

If you have any suggestions for improvements, or questions, please also raise an issue.

Citing the Package

If you use the VineCopulas package in your work, please cite:

Claassen, J. N., Koks, E. E., de Ruiter, M. C., Ward, P. J., & Jäger, W. S. (2024). VineCopulas: an open-source Python package for vine copula modelling. Journal of Open Source Software, 9(101), 6728. https://doi.org/10.21105/joss.06728