Clustering of terrorism deaths dataset before and after 2001
Dominika Lach
2024-01-15
Introduction to the topic and data set
The terrorist attack on the United States carried out on September 11, 2001 reverberated around the world. The 9/11 attacks caused far-reaching political, economic and social repercussions, thus leaving a mark on the culture not only of the United States, but of the entire West. Perhaps the most significant effect of the 2001 attacks was the so-called “war on terror” that the United States launched.
Project rely on the Global Terrorism Database (GTD) which can be found here: https://ourworldindata.org/terrorism?insight=terrorism-is-a-rare-cause-of-death-globally#key-insights
The GTD defines a terrorist attack as the threat or use of violence to achieve a political, economic, religious, or social goal through intimidation or coercion by a non-state actor. Because the data is based on news reports, attacks, and deaths may be underreported in countries with sparser media, especially during early years.
Data set contains data from 1970 till 2021. We will focus on comparison how the number of victims caused by terrorism differs in these two periods.
source: ourworldindata.org
Line plot with whole world data showing amount of terrorism deaths during the years
Terrorism map since 11 September 2001
source: wyborcza.pl
Comparison of terrorism sum of terrorism deaths by regions before and after 2001
It is very evident that the number of deaths caused by terrorism has increased after 2001. This is a result of both the “war on terror” and more modern surveying and military technology.
Clustering using k-means
The purpose of the clustering is to identify countries and the highest number of fatalities both before and after 2001, and to compare the results.
Before 2001
Normalization of data and checking what is the optimal number of clusters
# Normalization of data
Countries_before_norm <- scale(Countries_Before2001[,c('Year','Terrorism.deaths')])
opt1_before<-Optimal_Clusters_KMeans(Countries_before_norm, max_clusters=10, plot_clusters = TRUE)
Clustering
# Clustering with 4 clusters
fviz_cluster(kmeans_flex_before,palette=my_palette2)
n_clusters_before <- 4
kmeans_result_before <- kmeans(Countries_before_norm, centers = n_clusters_before)
# assitgning clusters to Countries
Countries_Before2001$Cluster <- as.factor(kmeans_result_before$cluster)
palette(my_palette2)
plot(Countries_before_norm, col = Countries_Before2001$Cluster, pch = 16, main = "Clustering results", xlab = "Year", ylab = "Terrorism deaths")
legend("topleft", legend = levels(Countries_Before2001$Cluster), col = 1:n_clusters_before, pch = 16, title = "Cluster")
Let’s check what data is in cluster 1
cluster_top_before <- Countries_Before2001[Countries_Before2001$Cluster == 1,]
cluster_top_before <- cluster_top_before[order(cluster_top_before$Terrorism.deaths, decreasing = TRUE),]
cluster_top_before## Entity Code Year Terrorism.deaths Cluster
## 180 Algeria DZA 1997 4266 1
## 6592 Nicaragua NIC 1984 3617 1
## 6591 Nicaragua NIC 1983 3277 1
## 9648 United States USA 2001 3014 1
## 7204 Peru PER 1984 2467 1
## 2884 El Salvador SLV 1980 2383 1
## 8658 Sri Lanka LKA 1989 1822 1
## 7203 Peru PER 1983 1818 1
## 2887 El Salvador SLV 1983 1667 1
## 2885 El Salvador SLV 1981 1578 1
## 6593 Nicaragua NIC 1985 1575 1
## 7651 Rwanda RWA 1994 1571 1
## 3872 Guatemala GTM 1982 1562 1
## 2886 El Salvador SLV 1982 1549 1
## 7209 Peru PER 1989 1502 1
## 7211 Peru PER 1991 1412 1
## 8659 Sri Lanka LKA 1990 1393 1
## 3871 Guatemala GTM 1981 1381 1
## 1471 Burundi BDI 1996 1371 1
## 2888 El Salvador SLV 1984 1315 1
## 8656 Sri Lanka LKA 1987 1274 1
## 9354 Turkey TUR 1992 1233 1
## 1982 Colombia COL 1997 1230 1
## 4341 India IND 1992 1152 1
## 7210 Peru PER 1990 1146 1
## 4340 India IND 1991 1113 1
## 1974 Colombia COL 1988 1096 1
## 8660 Sri Lanka LKA 1991 1034 1
## 7654 Rwanda RWA 1997 1027 1
## 8663 Sri Lanka LKA 1995 1020 1
## 2889 El Salvador SLV 1985 1013 1
## 8661 Sri Lanka LKA 1992 984 1
## 9355 Turkey TUR 1994 982 1
## 4337 India IND 1988 966 1
## 181 Algeria DZA 1998 952 1
## 1971 Colombia COL 1985 935 1
## 4339 India IND 1990 907 1
## 8657 Sri Lanka LKA 1988 885 1
## 4338 India IND 1989 874 1
## 7208 Peru PER 1988 855 1
## 4345 India IND 1997 853 1
## 286 Angola AGO 2001 846 1
## 8664 Sri Lanka LKA 1996 827 1
## 7212 Peru PER 1992 818 1
## 7207 Peru PER 1987 773 1
## 1972 Colombia COL 1986 771 1
## 1975 Colombia COL 1989 751 1
## 1977 Colombia COL 1991 742 1
## 7205 Peru PER 1985 729 1
## 6959 Pakistan PAK 1995 712 1
## 6189 Mozambique MOZ 1989 710 1
## 1976 Colombia COL 1990 694 1
## 8655 Sri Lanka LKA 1986 690 1Between 1970 and 2001, the highest number of deaths caused by terrorist attacks was in 1997 in Algeria, where the number of cases was 4266 people.
After 2001
Normalization of data and checking what is the optimal number of clusters
Countries_after_norm <- scale(Countries_After2001[,c('Year','Terrorism.deaths')])
opt1_after<-Optimal_Clusters_KMeans(Countries_after_norm, max_clusters=10, plot_clusters = TRUE)
Clustering
fviz_cluster(kmeans_flex_after,palette=my_palette2)
n_clusters_after <- 4
kmeans_result_after <- kmeans(Countries_after_norm, centers = n_clusters_after)
# assitgning clusters to Countries
Countries_After2001$Cluster <- as.factor(kmeans_result_after$cluster)
palette(my_palette2)
plot(Countries_after_norm, col = Countries_After2001$Cluster, pch = 16, main = "Clustering results", xlab = "Year", ylab = "Terrorism deaths")
# Legenda
legend("topleft", legend = levels(Countries_After2001$Cluster), col = 1:n_clusters_after, pch = 16, title = "Cluster")
Let’s check what data is in cluster 4
cluster_after <- Countries_After2001[Countries_After2001$Cluster == 4,]
cluster_after <- cluster_after[order(cluster_after$Terrorism.deaths, decreasing = TRUE),]
cluster_after## Entity Code Year Terrorism.deaths Cluster
## 4566 Iraq IRQ 2014 14095 4
## 4568 Iraq IRQ 2016 12276 4
## 50 Afghanistan AFG 2020 10081 4
## 48 Afghanistan AFG 2018 9812 4
## 4567 Iraq IRQ 2015 8885 4
## 49 Afghanistan AFG 2019 8257 4
## 6723 Nigeria NGA 2014 7775 4
## 4565 Iraq IRQ 2013 7042 4
## 4559 Iraq IRQ 2007 6665 4
## 4569 Iraq IRQ 2017 6619 4
## 45 Afghanistan AFG 2015 6216 4
## 46 Afghanistan AFG 2016 6141 4
## 47 Afghanistan AFG 2017 6095 4
## 6724 Nigeria NGA 2015 5567 4
## 44 Afghanistan AFG 2014 5414 4
## 4558 Iraq IRQ 2006 4612 4
## 51 Afghanistan AFG 2021 4337 4
## 8989 Syria SYR 2015 3924 4
## 43 Afghanistan AFG 2013 3709 4Between 2001 and 2021, the highest number of deaths caused by terrorist attacks was in 2014 in Iraq, where the number of cases was 14095 people.
Conclusion
There is a huge difference in the number of deaths before and after 2001. Note that after the landmark date, the countries with the highest number of deaths are primarily Iraq and Afghanistan. Since, according to the US, al-Qa’ida had its training bases precisely in Afghanistan, the George W. Bush government issued an ultimatum to the country’s ruling Taliban. They were to hand over Osama bin Laden to the Americans and shut down terrorist bases inside the country. The ultimatum was rejected, resulting in the outbreak of war in Afghanistan on October 7, 2001. It also involved NATO on the US side, which considered the 9/11 attacks a violation of Article 5 of the North Atlantic Alliance. The war in Afghanistan did not end until 2021, when the Joe Biden administration ordered the withdrawal of U.S. forces from the country. The Taliban took back power. The United States has also accused Iraq of supporting terrorism, with additional weapons of mass destruction at its disposal.
source: “wszystkoconajwazniejsze.pl”