ARmed COnflict avaLANCHEs (arcolanche)

This repository contains code for the construction and analysis of conflict avalanches, including simulation of the model and incorporation of social data.

If you use this Python package for your research, please cite the following paper:

Niraj Kushwaha , Edward D Lee, Discovering the mesoscale for chains of conflict, PNAS Nexus, Volume 2, Issue 7, July 2023, pgad228, https://doi.org/10.1093/pnasnexus/pgad228

Important modules

• construct.py: This module is used to construct conflict avalanches.
• network.py: This module is used to construct causal network using transfer entropy.
• pipeline.py: This file contains routines for the final analysis.

Installation

To run the notebook, please follow these steps:

1. Clone this specific branch of the repository and move inside the repository which will be our working directory:

   $ git clone --branch PNAS_Nexus_2023 https://github.com/eltrompetero/armed_conflict_avalanche.git
   $ cd armed_conflict_avalanche
   

2. Install custom repositories into your current working directory:

   $ git clone https://github.com/eltrompetero/misc.git
   $ git clone https://github.com/eltrompetero/workspace.git
   

3. Install the custom repository Voronoi_globe as per the instructions given here.

4. Install a custom Anaconda virtual environment using the spec file provided in the repository. To do that, inside the working directory (which contains the spec file), run the following command:

   $ conda create --name <ENVIRONMENT NAME> --file specfile_armed_conflict.txt
   

5. Install vincenty and mycolorpy using pip in the conda environment since they are unavailable in conda-forge.

    $ conda activate <ENVIRONMENT NAME>
    $ python3 -m pip install vincenty
    $ python3 -m pip install mycolorpy
   

   If some packages are missing, it is easiest to install them from the conda-forge channel. For example, if geopandas is missing, run the following command:

   $ conda install -n <ENVIRONMENT NAME> -c conda-forge geopandas
   

6. Copy the voronoi_grids and avalanches folder to the working directory. You can download these folder from Zenodo. If you wish to generate new voronoi grids, follow the instructions at Voronoi_globe.

7. Create a new folder called data and add your ACLED dataset to this folder. You can download a filtered version (only necessary information corresponding to each conflict event is kept) of the ACLED dataset that we used in our analysis from here. If you wish to download the full version please download it from ACLED. Make sure that the datafile is renamed as ACLED_data.csv. We used the following parameters to download the dataset from ACLED (we downloaded the dataset on 30th September 2022):

   From : 01/01/1997
   To : 31/12/2019
   Event Type : All
   Sub Event : All
   Actor Type : All
   Actor : All
   Region : Eastern Africa, Middle Africa, Northern Africa, Southern Africa, Western Africa
   Country : All
   Location : All
   

You can use these filters to get the same dataset as us.

Note: If you want to use this code to analyse regions other than Africa, you need to first generate voronoi grids for that region (Use Voronoi_globe).

Reproducing figures used in the paper

1. Inside the working directory, activate the custom conda environment.

   $ conda activate <ENVIRONMENT NAME>
   

2. Run paper_pipeline.ipynb and go through the instructions.

Questions and Feedback

If you have any questions about how to utilize specific features of our Python package or if you would like to suggest improvements to the code, we encourage you to reach out. We value your feedback and are committed to enhancing your experience with our package.

Please feel free to use the discussions and issues section of github for any inquiries or suggestions you may have. We appreciate your contribution to making our package even better.

Thank you for using our Python package!