African Armed Conflicts: Visualization & Forecasting

Visualizing Interstate and Internal Conflicts using Small & Medium Arms Imports, Military Expenditures, Fraction of Population in the Army, etc. We use a time-series framework to tackle the problem.

Link to interactive map

Model

Bagged Random Forest (0.9) and Neural Network (0.1)

Overview

Let's first look at all the files in the repo

$ tree --dirsfirst --filelimit 100

.
├── analysis
│   ├── data
│   │   ├── nisat
│   │   │   ├── resources
│   │   │   │   ├── NISAT database public user manual.pdf
│   │   │   │   └── nisat_query.png
│   │   │   ├── arms_trades_exports.csv
│   │   │   ├── arms_trades_imports.csv
│   │   │   ├── arms_trades.zip
│   │   │   ├── preproc_arms_import.csv
│   │   │   └── prio_weapons_code.txt
│   │   ├── output
│   │   │   ├── african_countries.json
│   │   │   ├── arms_imports.json
│   │   │   ├── conflicts.json
│   │   │   ├── mil_exp.json
│   │   │   ├── mil_pers.json
│   │   │   ├── population.json
│   │   │   └── predictions.json
│   │   ├── owid
│   │   │   ├── military-expenditure-as-a-share-of-gdp.csv
│   │   │   └── military-personnel-relative-to-total-population.csv
│   │   ├── population
│   │   │   └── world_population.csv
│   │   ├── ucdp
│   │   │   ├── resources
│   │   │   │   └── ucdp-dyadic-191.pdf
│   │   │   └── ucdp-dyadic-191.csv
│   │   └── world
│   │       ├── african_countries.csv
│   │       ├── country_continent.csv
│   │       └── COW country codes.csv
│   ├── african_countries_dict.ipynb
│   ├── nisat-1-scraping.ipynb
│   ├── owid-1-milexp.ipynb
│   ├── owid-1-milpers.ipynb
│   ├── population.ipynb
│   ├── regression.ipynb
│   └── ucdp-1-preproc.ipynb
├── dataviz
│   ├── assets
│   │   ├── css
│   │   │   ├── leaflet.css
│   │   │   └── style.css
│   │   ├── img
│   │   │   └── search.svg
│   │   └── libs
│   │       ├── d3.js
│   │       ├── d3-tip.js
│   │       ├── leaflet.js
│   │       └── localization-en.js
│   ├── data
│   │   ├── african_countries.json
│   │   ├── arms_imports.json
│   │   ├── conflicts.json
│   │   ├── mil_exp.json
│   │   ├── mil_pers.json
│   │   ├── population.json
│   │   ├── predictions.json
│   │   └── world.json
│   ├── scripts
│   │   ├── 1-preproc.js
│   │   ├── 2-map.js
│   │   └── main.js
│   └── index.html
├── README.md
└── screenshot.png

analysis/ : Contains the notebooks to collect, process, train and predict

dataviz/ : Contains the interactive map, made with the D3.js library

Datasets

• Norwegian Initiative on Small Arms Transfers
• Uppsala Conflict Data Program
• Our World in Data
• The World Bank

Launching jupyter notebook

Make sure to cd in analysis/ before launching jupyter notebook. If not, the paths won't work

Authors

• Jean-Romain Roy - Co-author: data collection, preprocessing, logistic regression, random forest, svm, interactive map - jeanromainroy