Event types that result in negative health and economic consequences from 1950-2011

Synopsis

The data analyzes both health and economic consequences by event type on US counties from 1950-2011. Starting with the health consequences, we can divide the data into deaths and injuries. The data shows that the event type that has caused the greatest harm (in both casualties and injuries) has been tornados. On the contrary, the greatest economic harm has not been caused by tornados, but has been dominated by hail.

Data processing

The data was loaded into R by downloading the file from the internet and read using the “.csv.bz2” extension.

if(!exists("P2")){
download.file(url = "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2", destfile = "P2.csv.bz2", method ="curl")
P2 <-  read.csv("P2.csv.bz2")
}

Results

options(scipen=999)
fatal <- with(P2,aggregate(FATALITIES, by = list(EVTYPE), FUN = sum))
max_fatal <- tolower(as.character(fatal[which.max(fatal$x),]$Group.1))
x <- fatal$x
names(x) <- fatal$Group.1
fatal2 <- x[x > 300]
par(mfrow = c(2,1), mgp = c(5,1,0), oma = c(2,2,0,0))
barplot(fatal2, cex.names = 0.5, las = 2, space = 0, ylim = c(0,6000), main = "Cumulative number of fatalities from 1950-2011", yaxp = c(0, 6000, 3))
mtext(expression(italic("Only includes events with cum fatalities of 300+")), cex = 0.8)

injur <- with(P2,aggregate(INJURIES, by = list(EVTYPE), FUN = sum))
max_injury <- tolower(as.character(injur[which.max(injur$x),]$Group.1))
y <- injur$x
names(y) <- injur$Group.1
injur2 <- y[y > 6000]
barplot(injur2, cex.names = 0.5, las = 2, space = 0, ylim = c(0,100000), main = "Cumulative number of injuries from 1950-2011", yaxp = c(0, 100000, 2))
mtext(expression(italic("Only includes events with cum injuries of 6k+")), cex = 0.8)

The most fatal event type was tornado
The most harmful (i.e. injury) event type was tornado

options(scipen=999)
par(mfrow = c(1,1), mar = c(7,6,4,2), oma = c(2,2,0,0), mgp = c(5,1,0))
economic <- with(P2,aggregate(CROPDMG, by = list(EVTYPE), FUN = sum))
max_econ <- tolower(as.character(economic[which.max(economic$x),]$Group.1))
z <- economic$x
names(z) <- injur$Group.1
economic2 <- z[z > 100000]
barplot(economic2, cex.names = 0.5, las = 2, space = 0, ylim = c(0,600000), main = "Cumulative number of crop damage from 1950-2011", xlab = "Event type", ylab = "Cumulative crop damage")
mtext(expression(italic("Only includes events with cum dmg of 100k+")), cex = 0.8)

The worst economic event type was hail

library("xtable")
library("plyr")
fat_sum <- with(P2,aggregate(FATALITIES,list(STATE),sum))
inj_sum <- with(P2,aggregate(INJURIES,list(STATE),sum))
eco_sum <- with(P2,aggregate(CROPDMG,list(STATE),sum))
all_sum <- join_all(list(fat_sum,inj_sum,eco_sum), by = "Group.1")
names(all_sum) <- c("State", "Fatalities", "Injuries", "Econdmg")
new_sum <- all_sum[(all_sum$Fatalities > quantile(all_sum$Fatalities)[["75%"]]) & (all_sum$Injuries > quantile(all_sum$Injuries)[["75%"]]) & (all_sum$Econdmg > quantile(all_sum$Econdmg)[["75%"]]),]
sum_row <- rbind(summary(fat_sum$x),summary(inj_sum$x),summary(eco_sum$x))
row.names(sum_row) <- c("Fatalities", "Injuries", "Econ dmg")
print(xtable(sum_row, digits = 0), type = "html")

	1st Qu.	Median	Mean	3rd Qu.	Max.
Fatalities	12	86	210	314	1421
Injuries	44	520	1952	3217	17667
Econ dmg	24	3490	19136	15813	303950

Here, in the table above we see a distribution of fatalities, injuries and econ damage numbers

print(xtable(new_sum, digits = 0), type = "html")

	State	Fatalities	Injuries	Econdmg
5	AR	530	5550	25819
8	CA	550	3278	21152
22	KS	356	3449	139799
37	MO	754	8998	21087
38	MS	555	6675	56078
40	NC	398	3415	22246
48	OH	403	7112	28328
63	TX	1366	17667	156169

Here above we see in the table the 8 states which had fatalities, injuries and economic danage in the highest quantile (i.e. worst economic and health consequences)