Spatial Analysis of ACLED Data

R tool for geographically analysing ACLED data.

ACLED

The Armed Conflict Location & Event Data Project (ACLED) is a disaggregated data collection, analysis, and crisis mapping project ACLED collects real-time data on the locations, dates, actors, fatalities, and types of all reported political violence and protest events around the world.

Repository structure

• Installations
• Setting Environment
• Data
• Cleaning & Feature Engineering
• Visualization
  • Part 1: Summary Data
  • Part 2: Spatial Analysis

Installations

Install the latest release version of required packages from CRAN with :

install.packages(c('acled.api','dplyr', 'tibble', 'janitor', 'plotly', 'ggplot2'))

install.packages(c("sf", "leaflet", "leaflet.providers", "leaflet.opacity", "cartography"))

Setting Environment

Call the library functions from the installed packages as follows:

ACLED API

library(acled.api)

Tools for data manipulation feature engineering:

library(dplyr)
library(tibble)
library(janitor)

Summary visuals

library(plotly)
library(ggplot2)
library(RColorBrewer)

Spatial Analysis and Visuals

library(sf)
library(tmap)
library(leaflet)
library(tmaptools)
library(cartography)
library(leaflet.opacity)
library(leaflet.providers)

Plotly Image Export

Using Plotly R package for summary visuals we would also want to save the static images. Set the environment variables in the R session using Sys.setenv().

Sys.setenv("plotly_username" = "Musebe")
Sys.setenv("plotly_api_key" = "*********")

Data

Export ACLED data to R usin the ACLED API:

Crime_in_Zimababwe <- acled.api(
  email.address = "xxxx",
  access.key = "xxxx",
  country = "Zimbabwe",
  start.date = "2005-01-01",
  end.date = "2021-07-01",
  add.variables = c("latitude","longitude", "geo_precision", "time_precision")

The above function imitates the ACLED data export tool. The two credentials required for the API to function are the email adress and the access key which are availed upon subcribing to the platform.

Sample view of the data:

iso3	year	event_date	source	admin1	admin2	location	event_type	sub_event_type	interaction	fatalities	timestamp	latitude	longitude	geo_precision	time_precision
ZWE	2021	28-Jun-21	Zimbabwean	Matabeleland North	Hwange Urban	Hwange	Protests	Protest with intervention	16	0	1625510722	-18.3693	26.5019	2	2
ZWE	2021	18-Jun-21	Zim Eye	Harare	Harare Urban	Harare	Violence against civilians	Attack	17	0	1625510722	-17.8277	31.0534	1	1
ZWE	2021	16-Jun-21	New Zimbabwe	Mashonaland West	Kadoma Urban	Kadoma	Protests	Peaceful protest	60	0	1625510721	-18.35	29.9167	1	1
ZWE	2021	9-Jun-21	Bulawayo24	Bulawayo	Bulawayo	Bulawayo	Riots	Mob violence	15	0	1624310472	-20.15	28.58	1	1
ZWE	2021	7-Jun-21	Chronicle (Zimbabwe)	Bulawayo	Bulawayo	Bulawayo	Strategic developments	Looting/property destruction	60	0	1624310472	-20.15	28.58	1	1

Cleaning & Feature Engineering

Adding a column time past since 1979-01-01. Also for purpose of summarizing data in the year create the day of the week and month variable. Both of them from the event_date variable.

NB: Little cleaning is required for the ACLED data as most of the data is already sorted and ready for analysis.

Crime_in_Zimbabwe <- Crime_in_Zimbabwe %>% 
  add_column(
    # time
    time = format(as.POSIXct(
      Crime_in_Zimbabwe$timestamp, origin="1970-01-01"), format = "%H:%M:%S"),
    # day of the week
    dow = weekdays(as.Date(Crime_in_Zimbabwe$event_date), abbreviate = T),
    #Month of the Year
    month = months(as.Date(Crime_in_Zimbabwe$event_date), abbreviate = TRUE)
  )

Visualization

Part 1: Summary Data

Sample visuals extracted from the data are as follows:

Top Ten location with highest crisis rate

Harare faces high crime activity than other locations in the country.

Provincial crisis rate

Harare province has the highest crisis rate. This is relative with Harare town being in the same province.

Monthly Crime Rate

Distribution of crime activity:

Crime Rate by sub_event

Crime Rate Against Male Population

The visual suggests negative correlation between male population and the crime rate in all the provinces. An outlier is seen un the Harare province. This

Part 2: Spatial Analysis

Geneat visuals of the administrative boundaries of Zimbabwe can be produced by:

leaflet() %>%
  addProviderTiles(providers$OpenStreetMap.HOT) %>%
  addPolygons(data = level_3 , # borders of all wards
              color = "grey", 
              fill = NA,
              weight = 0.8) %>%
  addPolygons(data = level_2 , # borders of all districts
              color = "blue", 
              fill = NA,
              weight = 1) %>%
  addPolygons(data = level_1 , # borders of all provinces
              color = "red", 
              fill = NA,
              weight = 1.5) %>%
  addPolygons(data = zim , # borders of the Country
              color = "grey", 
              fill = NA,
              weight = 4)

Example:

Provicial Crime Rate

By leaflet package(left image) & cartography package:

Copyright and license

The sample data used here belongs to and has been extracted from ACLED.