Model terrorism patterns in the United Kingdom

terrorism-in-ukg-1970-2015.R

# You won't need all of these packages. I just copied it from another,related analysis:
# https://github.com/svmiller/etjc

library(car)
library(arm)
library(countrycode)
library(data.table)
library(tidyverse)
library(sqldf)
# library(pdftools)
library(stringr)
library(labelled)
library(countrycode)
library(WDI)
library(mirt)

GTD93 <- readxl::read_excel("~/Dropbox/data/gtd/20160629-download/gtd1993_0616dist.xlsx")
GTD <- readxl::read_excel("~/Dropbox/data/gtd/20160629-download/globalterrorismdb_0616dist.xlsx")


States <- read_csv("~/Dropbox/data/MID/states/states2011.csv")

GTD$ccode <- countrycode(GTD$country_txt, "country.name", "cown")
GTD93$ccode <- countrycode(GTD93$country_txt, "country.name", "cown")

GTD %>% 
  mutate(intnl = ifelse(INT_LOG == 1 | INT_IDEO == 1 | INT_MISC == 1, 1, 0),
         pubtarg = ifelse(targtype1 == 1 | targtype1 == 5 | targtype1 == 6 | targtype1 == 8 |
                            targtype1 == 9 | targtype1 == 14 | targtype1 == 15 | 
                            targtype1 == 15 | targtype1 == 16 | targtype1 == 18 |
                            targtype1 == 19, 1, 0)) %>%
  select(ccode, country_txt, iyear, imonth, iday,
         success, nkill, nwound, propextent, intnl, pubtarg) -> GTDs

GTD93 %>%
  mutate(intnl = NA,
         pubtarg =  ifelse(targtype1 == 1 | targtype1 == 5 | targtype1 == 6 | targtype1 == 8 |
                             targtype1 == 9 | targtype1 == 14 | targtype1 == 15 | 
                             targtype1 == 15 | targtype1 == 16 | targtype1 == 18 |
                             targtype1 == 19, 1, 0)) %>%
  select(ccode, country_txt, iyear, imonth, iday,
         success, nkill, nwound, propextent, intnl, pubtarg) %>%
  bind_rows(., GTDs) %>%
  arrange(iyear, imonth, iday, ccode) %>%
  rename(year = iyear, month = imonth, day = iday) -> GTDf

GTDf %>%
  mutate(dependency = ifelse(is.na(ccode), 1, 0),
         ccode = ifelse(country_txt == "West Germany (FRG)", 260, ccode),
         ccode = ifelse(country_txt == "West Bank and Gaza Strip", 666, ccode),
         ccode = ifelse(country_txt == "Guadeloupe", 220, ccode),
         ccode = ifelse(country_txt == "Martinique", 220, ccode),
         ccode = ifelse(country_txt == "New Caledonia", 220, ccode),
         ccode = ifelse(country_txt == "Northern Ireland", 200, ccode),
         # Fortunately, no HK terror incidents in 1997
         ccode = ifelse((country_txt == "Hong Kong" & year < 1998), 200, ccode),
         ccode = ifelse((country_txt == "Hong Kong" & year > 1997), 710, ccode),
         # Fortunately, no Macau terror incidents in 1999
         ccode = ifelse((country_txt == "Macau" & year < 2000), 235, ccode),
         ccode = ifelse((country_txt == "Serbia" | 
                           country_txt == "Serbia-Montenegro"), 345, ccode),
         ccode = ifelse(country_txt == "French Guiana", 220, ccode),
         ccode = ifelse(country_txt == "Corsica", 220, ccode),
         ccode = ifelse(country_txt == "Gibraltar", 200, ccode),
         ccode = ifelse(country_txt == "Falkland Islands", 200, ccode),
         ccode = ifelse(country_txt == "French Polynesia", 220, ccode),
         ccode = ifelse(country_txt == "Wallis and Futuna", 220, ccode),
         dependency = ifelse(country_txt == "Northern Ireland" | ccode == 345 | ccode == 710, 0, dependency)
  ) -> GTDf

# Some recoding decisions ----------
# success: if we don't know if it was successful, it wasn't.
# nkill: if it's missing, we'll say 1. People died, but we'll be conservative.
# nwound: if it's missing, we'll say 1. People were injured, but we'll be conservative.
# propextent: if it's missing, code it as zero. This is consistent w/ GTI.

GTDf %>%
  mutate(ones = 1,
         success = ifelse(is.na(success), 0, success),
         nkill = ifelse(is.na(nkill), 1, nkill),
         nwound = ifelse(is.na(nwound), 1, nwound),
         propextent = ifelse(is.na(propextent), 0, propextent)) %>%
  # Might as well summarize now...
  group_by(year, ccode) %>%
  summarize(ones = sum(ones),
            success = sum(success),
            nkill = sum(nkill),
            nwound = sum(nwound),
            propextent = sum(propextent)) -> GTDf

# Side track to create a PDF of state-years. ----------
# States data nominally end in 2011, so let's fudge that.

States %>%
  mutate(endyear = 2015) %>%
  rowwise() %>%
  mutate(year = list(seq(styear, endyear))) %>%
  ungroup() %>%
  unnest() %>%
  arrange(ccode, year) %>%
  select(ccode, year) %>%
  filter(year >= 1970) %>%
  distinct(ccode, year) -> CYs

CYs %>% 
  group_by(ccode, year) %>% 
  filter(n()>1)

CYs <- left_join(CYs, GTDf) %>%
  mutate(ones = ifelse(is.na(ones), 0, ones),
         success = ifelse(is.na(success), 0, success),
         nkill = ifelse(is.na(nkill), 0, nkill),
         nwound = ifelse(is.na(nwound), 0, nwound),
         propextent = ifelse(is.na(propextent), 0, propextent))

CYs %>%
  group_by(ccode) %>%
  mutate(gtiraw = ones + 3*(nkill) + .5*(nwound) + 2*(propextent),
         l1_gtiraw = lag(gtiraw, 1),
         l2_gtiraw = lag(gtiraw, 2),
         l3_gtiraw = lag(gtiraw, 3),
         l4_gtiraw = lag(gtiraw, 4),
         l5_gtiraw = lag(gtiraw, 5),
         gtiraw5ya = 16*(l1_gtiraw) + 8*(l2_gtiraw) + 4*(l3_gtiraw) + 2*(l4_gtiraw) + l5_gtiraw,
         loggti5ya = log(gtiraw5ya + 1)) %>%
  group_by(year) %>%
  mutate(#maxgtiraw = max(gtiraw5ya, na.rm=T),
    #gtilogbase = maxgtiraw^(1/20),
    loggti5ya = log(gtiraw5ya + 1, base=max(gtiraw5ya, na.rm=T)^(1/20))
  ) %>%
  select(-l1_gtiraw, -l2_gtiraw, -l3_gtiraw, -l4_gtiraw, -l5_gtiraw) -> CYs


CYs %>%
  filter(ccode == 200) -> UKG


theme_steve <- function() {
  theme_bw() +
    theme(panel.border = element_blank(), 
          plot.caption=element_text(hjust=1, size=9,
                                    margin=margin(t=10),
                                    face="italic"),
          plot.title=element_text(hjust=0, size=18,
                                  margin=margin(b=10),
                                  face="bold"),
          axis.title.y=element_text(size=12,hjust=1,
                                    face="italic"),
          axis.title.x=element_text(hjust=1, size=12, face="italic"))
  
}

UKG %>%
  ggplot(.,aes(year,ones)) + geom_line() + theme_steve() +
  xlab("Year") + ylab("Number of Terror Incidents") +
  annotate("rect",xmin=1970,xmax=1998, ymin=0, ymax=300, alpha=.3, fill="red") +
  annotate("text", x=1980, y=50, label="The Troubles\n(1968-1998)") +
  annotate("text", x=2007, y=200, label="Post-9/11\n(2001-2015)") +
  geom_segment(aes(x = 2001, y=0, xend=2001, yend=300), linetype="dashed", size=0.3) +
  scale_x_continuous(breaks=c(seq(1970, 2015, 5))) +
  ggtitle("Number of Terrorism Incidents in the United Kingdom (1970-2015)") +
  labs(subtitle = "Data come from June 2016 update of Global Terrorism Database")


UKG %>%
  ggplot(.,aes(year,success)) + geom_line() + theme_steve() +
  xlab("Year") + ylab("Number of Successful Terror Incidents") +
  annotate("rect",xmin=1970,xmax=1998, ymin=0, ymax=300, alpha=.3, fill="red") +
  annotate("text", x=1980, y=50, label="The Troubles\n(1968-1998)") +
  annotate("text", x=2007, y=200, label="Post-9/11\n(2001-2015)") +
  geom_segment(aes(x = 2001, y=0, xend=2001, yend=300), linetype="dashed", size=0.3) +
  scale_x_continuous(breaks=c(seq(1970, 2015, 5))) +
  ggtitle("Number of Successful Terrorism Incidents in the United Kingdom (1970-2015)") +
  labs(subtitle = "Data come from June 2016 update of Global Terrorism Database")

UKG %>%
  ggplot(.,aes(year,nkill)) + geom_line() + theme_steve() +
  xlab("Year") + ylab("Number of People Killed in Terror Incidents") +
  annotate("rect",xmin=1970,xmax=1998, ymin=0, ymax=400, alpha=.3, fill="red") +
  annotate("text", x=1980, y=50, label="The Troubles\n(1968-1998)") +
  annotate("text", x=2007, y=200, label="Post-9/11\n(2001-2015)") +
  geom_segment(aes(x = 2001, y=0, xend=2001, yend=400), linetype="dashed", size=0.3) +
  scale_x_continuous(breaks=c(seq(1970, 2015, 5))) +
  ggtitle("Number of People Killed in Terrorism Incidents in the United Kingdom (1970-2015)") +
  labs(subtitle = "Data come from June 2016 update of Global Terrorism Database")

UKG %>%
  ggplot(.,aes(year,nkill)) + geom_line() + theme_steve() +
  xlab("Year") + ylab("Number of Terror Incidents") +
  annotate("rect",xmin=1970,xmax=1998, ymin=0, ymax=400, alpha=.3, fill="red") +
  annotate("text", x=1980, y=50, label="The Troubles\n(1968-1998)") +
  annotate("text", x=2007, y=200, label="Post-9/11\n(2001-2015)") +
  geom_segment(aes(x = 2001, y=0, xend=2001, yend=400), linetype="dashed", size=0.3) +
  scale_x_continuous(breaks=c(seq(1970, 2015, 5))) +
  ggtitle("Number of People Killed in Terrorism Incidents in the United Kingdom (1970-2015)") +
  labs(subtitle = "Data come from June 2016 update of Global Terrorism Database")