Global Terrorism Database Analysis
Marie Daras LLC
2025-03-20
Marie Daras LLC
Global Terrorism Database Analysis
Counter Terrorism
These are EDA’s and an analysis run on the Global Terrorism Database (GTD), which is publicly available and kept by the University of Maryland: GTD Univ of MD
Here is the codebook for the dataset: GTD Codebook
This is our Roadmap for our analysis:
Roadmap
library(dplyr) # for data manipulation
library(stringr) # for data manipulation
library(ggplot2) # for data visualization
library(corrplot) # for correlations
library(ggplot2)
library(RColorBrewer)
library(nnet)
library(tidyr)
library(plyr)
library(leaflet)
library(sf)
library(tidyverse)1) Bring in Our Data
df <- read.csv("/cloud/project/gtd.csv", header=TRUE, stringsAsFactors=FALSE)
head(df,1)## eventid iyear imonth iday approxdate extended resolution country
## 1 1.97e+11 1970 7 2 0 58
## country_txt region region_txt provstate city
## 1 Dominican Republic 2 Central America & Caribbean Santo Domingo
## latitude longitude specificity vicinity location summary crit1 crit2 crit3
## 1 18.45679 -69.95116 1 0 1 1 1
## doubtterr alternative alternative_txt multiple success suicide attacktype1
## 1 0 NA 0 1 0 1
## attacktype1_txt attacktype2 attacktype2_txt attacktype3 attacktype3_txt
## 1 Assassination NA NA
## targtype1 targtype1_txt targsubtype1 targsubtype1_txt corp1
## 1 14 Private Citizens & Property 68 Named Civilian
## target1 natlty1 natlty1_txt targtype2 targtype2_txt targsubtype2
## 1 Julio Guzman 58 Dominican Republic NA NA
## targsubtype2_txt corp2 target2 natlty2 natlty2_txt targtype3 targtype3_txt
## 1 NA NA
## targsubtype3 targsubtype3_txt corp3 target3 natlty3 natlty3_txt gname
## 1 NA NA MANO-D
## gsubname gname2 gsubname2 gname3 gsubname3 motive guncertain1 guncertain2
## 1 0 NA
## guncertain3 individual nperps nperpcap claimed claimmode claimmode_txt claim2
## 1 NA 0 NA NA NA NA NA
## claimmode2 claimmode2_txt claim3 claimmode3 claimmode3_txt compclaim
## 1 NA NA NA NA
## weaptype1 weaptype1_txt weapsubtype1 weapsubtype1_txt weaptype2 weaptype2_txt
## 1 13 Unknown NA NA
## weapsubtype2 weapsubtype2_txt weaptype3 weaptype3_txt weapsubtype3
## 1 NA NA NA
## weapsubtype3_txt weaptype4 weaptype4_txt weapsubtype4 weapsubtype4_txt
## 1 NA NA
## weapdetail nkill nkillus nkillter nwound nwoundus nwoundte property
## 1 1 NA NA 0 NA NA 0
## propextent propextent_txt propvalue propcomment ishostkid nhostkid nhostkidus
## 1 NA NA 0 NA NA
## nhours ndays divert kidhijcountry ransom ransomamt ransomamtus ransompaid
## 1 NA NA 0 NA NA NA
## ransompaidus ransomnote hostkidoutcome hostkidoutcome_txt nreleased addnotes
## 1 NA NA NA
## scite1 scite2 scite3 dbsource INT_LOG INT_IDEO INT_MISC INT_ANY related
## 1 PGIS 0 0 0 02) EDAs - Exploratory Data Analysis & Data Cleaning
Selecting only the variables that you wish to keep
After dropping the variables we don’t want these are the ones we are left with
Data Dictionary
df2 <-df %>% select(eventid, iyear,success,imonth, iday, country_txt,region_txt,suicide,latitude,longitude,attacktype1_txt, targtype1_txt, target1,nkill )
head(df2,1)## eventid iyear success imonth iday country_txt
## 1 1.97e+11 1970 1 7 2 Dominican Republic
## region_txt suicide latitude longitude attacktype1_txt
## 1 Central America & Caribbean 0 18.45679 -69.95116 Assassination
## targtype1_txt target1 nkill
## 1 Private Citizens & Property Julio Guzman 1Change eventid from numeric to character
df2$eventid <- as.factor(df2$eventid)Checking that it went well
class(df2$eventid)## [1] "factor"Datatypes for all vars
str(df2)## 'data.frame': 181691 obs. of 14 variables:
## $ eventid : Factor w/ 181691 levels "197000000001",..: 1 2 3 4 5 6 7 8 9 10 ...
## $ iyear : int 1970 1970 1970 1970 1970 1970 1970 1970 1970 1970 ...
## $ success : int 1 1 1 1 1 1 0 1 1 1 ...
## $ imonth : int 7 0 1 1 1 1 1 1 1 1 ...
## $ iday : int 2 0 0 0 0 1 2 2 2 3 ...
## $ country_txt : chr "Dominican Republic" "Mexico" "Philippines" "Greece" ...
## $ region_txt : chr "Central America & Caribbean" "North America" "Southeast Asia" "Western Europe" ...
## $ suicide : int 0 0 0 0 0 0 0 0 0 0 ...
## $ latitude : num 18.5 19.4 15.5 38 33.6 ...
## $ longitude : num -70 -99.1 120.6 23.8 130.4 ...
## $ attacktype1_txt: chr "Assassination" "Hostage Taking (Kidnapping)" "Assassination" "Bombing/Explosion" ...
## $ targtype1_txt : chr "Private Citizens & Property" "Government (Diplomatic)" "Journalists & Media" "Government (Diplomatic)" ...
## $ target1 : chr "Julio Guzman" "Nadine Chaval, daughter" "Employee" "U.S. Embassy" ...
## $ nkill : int 1 0 1 NA NA 0 0 0 0 0 ...EDAs
summary(df2)## eventid iyear success imonth
## 197000000001: 1 Min. :1970 Min. :0.0000 Min. : 0.000
## 197000000002: 1 1st Qu.:1991 1st Qu.:1.0000 1st Qu.: 4.000
## 197001000001: 1 Median :2009 Median :1.0000 Median : 6.000
## 197001000002: 1 Mean :2003 Mean :0.8896 Mean : 6.467
## 197001000003: 1 3rd Qu.:2014 3rd Qu.:1.0000 3rd Qu.: 9.000
## 197001010002: 1 Max. :2017 Max. :1.0000 Max. :12.000
## (Other) :181685
## iday country_txt region_txt suicide
## Min. : 0.00 Length:181691 Length:181691 Min. :0.00000
## 1st Qu.: 8.00 Class :character Class :character 1st Qu.:0.00000
## Median :15.00 Mode :character Mode :character Median :0.00000
## Mean :15.51 Mean :0.03651
## 3rd Qu.:23.00 3rd Qu.:0.00000
## Max. :31.00 Max. :1.00000
##
## latitude longitude attacktype1_txt targtype1_txt
## Min. :-53.16 Min. :-86185896 Length:181691 Length:181691
## 1st Qu.: 11.51 1st Qu.: 5 Class :character Class :character
## Median : 31.47 Median : 43 Mode :character Mode :character
## Mean : 23.50 Mean : -459
## 3rd Qu.: 34.69 3rd Qu.: 69
## Max. : 74.63 Max. : 179
## NA's :4556 NA's :4557
## target1 nkill
## Length:181691 Min. : 0.000
## Class :character 1st Qu.: 0.000
## Mode :character Median : 0.000
## Mean : 2.403
## 3rd Qu.: 2.000
## Max. :1570.000
## NA's :103133) Plot Terror Attacks By Region & Year
attacks <- df2 %>%
group_by(iyear, region_txt) %>%
tally()##Let’s look at number of terror attacks by year binned
region <- df2 %>%
group_by(region_txt) %>%
tally()print(region)## # A tibble: 12 × 2
## region_txt n
## <chr> <int>
## 1 Australasia & Oceania 282
## 2 Central America & Caribbean 10344
## 3 Central Asia 563
## 4 East Asia 802
## 5 Eastern Europe 5144
## 6 Middle East & North Africa 50474
## 7 North America 3456
## 8 South America 18978
## 9 South Asia 44974
## 10 Southeast Asia 12485
## 11 Sub-Saharan Africa 17550
## 12 Western Europe 16639Binning years by decade
df2$decade <- case_when(
df2$iyear %in% c("1970", "1971", "1972", "1973", "1974", "1975", "1976", "1977", "1978", "1979") ~ "1970's",
df2$iyear %in% c("1980", "1981", "1982", "1982", "1983", "1984", "1985", "1986", "1987", "1988", "1989") ~ "1980's",
df2$iyear %in% c("1990", "1991", "1992", "1992", "1993", "1994", "1995", "1996", "1997", "1998", "1999") ~ "1990's",
df2$iyear %in% c("2000", "2001", "2002", "2003", "2004", "2005", "2006", "2007", "2008", "2009") ~ "2000's",
df2$iyear %in% c("2010", "2011", "2012", "2013", "2014", "2015", "2016", "2017", "2018", "2019", "2020") ~ "2010's",
)library(ggplot2)
ggplot(attacks, aes(x = iyear, y = n, color = region_txt)) +
geom_line() 
The trendline shows a sharp increase in global terrorist attacks overall since after 2000 with a specific sharp increase after 2010.
4) Interactive Table of Attacks By Year & Regions
attacks2 <- df2 %>%
group_by(region_txt, decade, attacktype1_txt) %>%
tally()head(attacks2, 5)## # A tibble: 5 × 4
## # Groups: region_txt, decade [2]
## region_txt decade attacktype1_txt n
## <chr> <chr> <chr> <int>
## 1 Australasia & Oceania 1970's Armed Assault 1
## 2 Australasia & Oceania 1970's Assassination 6
## 3 Australasia & Oceania 1970's Bombing/Explosion 8
## 4 Australasia & Oceania 1970's Facility/Infrastructure Attack 1
## 5 Australasia & Oceania 1980's Armed Assault 15Rename Columns
#Rename columns
colnames(attacks2)[1] <- "Region"
colnames(attacks2)[2] <- "Decade"
colnames(attacks2)[3] <- "Attack Type"
head(attacks2, 5)## # A tibble: 5 × 4
## # Groups: Region, Decade [2]
## Region Decade `Attack Type` n
## <chr> <chr> <chr> <int>
## 1 Australasia & Oceania 1970's Armed Assault 1
## 2 Australasia & Oceania 1970's Assassination 6
## 3 Australasia & Oceania 1970's Bombing/Explosion 8
## 4 Australasia & Oceania 1970's Facility/Infrastructure Attack 1
## 5 Australasia & Oceania 1980's Armed Assault 15 attacks2 <- attacks2 %>%
mutate_if(is.character, utf8::utf8_encode)5) Map of Attacks By Region
Building our map dataframe
We’re doing it by region because country borders have changed as politics and wars have traversed by decade
We are going to use the libraries for mapping
library(tidyverse)
library(rnaturalearth)
library(sf)Get world data
world <- ne_countries(scale ="medium", returnclass="sf")%>%
filter(admin != "Anarctica")Build our dataframe to join
Terrorist Attacks Per Country/Region
terrorattacks <- df2 %>%
group_by(country_txt, decade) %>%
tally()Rename Columns for Joining
#Rename columns
colnames(terrorattacks)[1] <- "region"
colnames(terrorattacks)[2] <- "Decade"
colnames(terrorattacks)[3] <- "Freq"
head(terrorattacks, 5)## # A tibble: 5 × 3
## # Groups: region [1]
## region Decade Freq
## <chr> <chr> <int>
## 1 Afghanistan 1970's 4
## 2 Afghanistan 1980's 22
## 3 Afghanistan 1990's 98
## 4 Afghanistan 2000's 1949
## 5 Afghanistan 2010's 10658Add lat and long
latlong <- read.csv("/cloud/project/countrieslatlong.csv", header=TRUE, stringsAsFactors=FALSE)
head(latlong,1)## name lat long
## 1 Afghanistan 33 65Rename name to region
colnames(latlong)[1] <- "region"worldattacks <- merge(x=terrorattacks,y=latlong,
by="region", all.x=TRUE)Plot World Map
world_map <- map_data("world")Retrieve the world data and merge with terrorattacks data
terrorattacks_map <- left_join(world_map, worldattacks, by = "region")# Create the map
ggplot(terrorattacks_map, aes(long.x, lat.x, group = group ))+
geom_polygon(aes(fill = Freq), color = "lavender")+
scale_fill_viridis_c(option = "C")