A project for the Current Research Seminar WS22/23 by Christian Soraruf and Stefan Solarski at LMU.
The main file is a Jupyter notebook "Spatial Vine Copulas". It contains all of our outputs and visualizations.
The three main classes DataSets, SpatialCopula and SpatialVineCopula are located in the spvinecopulib.py file in the functions folder.
The main example dataset meuse can be loaded in the notebook using the skg package.
Clone the repository and use pip, or another package manager, to install the requirements.
git clone https://github.com/SSolarski/spatialvinecopula.git
cd spatialvinecopula
pip install -r requirements.txtNecessary packages are given in "requirements.txt".
The notebook can be run as is. (we used Python v3.8.8).
The alternative datasets from the SIC2004 exercise are saved as .csv files in the sic2004data_01 folder.
If you wish to run the algorithm on this dataset please uncomment the notebook cell which loads the optional dataset.
This work was based on ideas presented in the paper "Modelling skewed spatial random fields through the spatial vine copula" by Benjamin Gräler.
Our code is an attempt of recreating the functionalities of the R library spcopula by the same author.