Tornados and Floods are most dangerous weather events by their impact on public health and economy
Yevgeny V.Yorkhov
03/22/2015
Tornados and Floods have most dangerous impact on public health and economy
Synopsis
Severe weather events have bad impact on the human health and the economy. These events can result in fatalities, injuries, property and crops damages. As a result of the analysis we found that the most dangerous weather events are tornadoes and floods. In case of impact on human population the most dangerous weather events are tornadoes. The number of fatalities that tornadoes result are almost the same as the rest weather events combined. Also tornadoes result twice more injuries than the rest of other weather events combined. In case of economical impact floods are the most dangerous events. They have almost the same economical impact as all rest weather events combined. The most dangerous weather events for crops are droughts and floods.
Data processing
The data for this analysis come from National Weather Center in bzipped form. The archive can be downloaded from the following URL:
https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2
The events in the database start in the year 1950 and end in November 2011. In the earlier years of the database there are generally fewer events recorded, most likely due to a lack of good records. More recent years should be considered more complete.
For the analysis we used the following variables: - Event Type (EVTYPE) - Injuries and Fatalities (INJURIES, FATALITIES) - Property and crop damages (PROPDMG, CROPDMG) - Property and crop damage magnitude (PROPDMGEXP, CROPDMGEXP)
To calculate the financional loss in US Dollars we need to multiply PROPDMG and CROPDMG by 10 raised to the power of variables: PROPDMGEXP, CROPDMGEXP.
#url <- "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2"
#download.file(url, destfile = "repdata%2Fdata%2FStormData.csv.bz2")
dataset <- read.csv(bzfile("repdata%2Fdata%2FStormData.csv.bz2"))
df <- dataset[dataset$INJURIES!=0 | dataset$FATALITIES != 0 | dataset$PROPDMG != 0 | dataset$CROPDMG != 0, ]
## Clean dataset
## Work with levels
levels(df$CROPDMGEXP)[levels(df$CROPDMGEXP) == ""] <- 0
levels(df$CROPDMGEXP)[levels(df$CROPDMGEXP) == "?"] <- 0
levels(df$CROPDMGEXP)[levels(df$CROPDMGEXP) == "0"] <- 1
levels(df$CROPDMGEXP)[levels(df$CROPDMGEXP) == "2"] <- 100
levels(df$CROPDMGEXP)[levels(df$CROPDMGEXP) == "k"] <- 10^3
levels(df$CROPDMGEXP)[levels(df$CROPDMGEXP) == "K"] <- 10^3
levels(df$CROPDMGEXP)[levels(df$CROPDMGEXP) == "m"] <- 10^6
levels(df$CROPDMGEXP)[levels(df$CROPDMGEXP) == "M"] <- 10^6
levels(df$CROPDMGEXP)[levels(df$CROPDMGEXP) == "B"] <- 10^9
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "" | levels(df$PROPDMGEXP) == "?"] <- 0
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "+" | levels(df$PROPDMGEXP) == "-"] <- 0
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "1"] <- 10
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "0"] <- 1
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "2"] <- 100
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "h"] <- 100
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "H"] <- 100
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "3"] <- 1000
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "k"] <- 1000
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "K"] <- 1000
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "4"] <- 10^4
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "5"] <- 10^5
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "6"] <- 10^6
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "m"] <- 10^6
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "M"] <- 10^6
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "7"] <- 10^7
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "8"] <- 10^8
levels(df$PROPDMGEXP)[levels(df$PROPDMGEXP) == "B"] <- 10^9
df$CROPDMG_FIN <- df$CROPDMG*as.numeric(levels(df$CROPDMGEXP))[df$CROPDMGEXP]
df$PROPDMG_FIN <- df$PROPDMG*as.numeric(levels(df$PROPDMGEXP))[df$PROPDMGEXP]Results
Q1. Across the United States, which types of events (as indicated in the EVTYPE variable) are most harmful with respect to population health?
In order to answer this question we need to aggregate fatalities and injuries by event type. In both cases tornadoes are the most harmful weather events.
## Find sum of fatalities and injuries
dmg_hpop <- aggregate(cbind(FATALITIES, INJURIES) ~ EVTYPE, df, FUN=sum)
top10_fat<-head(dmg_hpop[order(dmg_hpop$FATALITIES, decreasing = T),1:2], n=10)
top10_inj<-head(dmg_hpop[order(dmg_hpop$INJURIES, decreasing = T),c(1,3)], n=10)Top 10 weather events that result huge fatalities
top10_fat## EVTYPE FATALITIES
## 405 TORNADO 5633
## 57 EXCESSIVE HEAT 1903
## 70 FLASH FLOOD 978
## 148 HEAT 937
## 252 LIGHTNING 816
## 423 TSTM WIND 504
## 83 FLOOD 470
## 304 RIP CURRENT 368
## 197 HIGH WIND 248
## 7 AVALANCHE 224Top 10 weather events that result huge injuries
top10_inj## EVTYPE INJURIES
## 405 TORNADO 91346
## 423 TSTM WIND 6957
## 83 FLOOD 6789
## 57 EXCESSIVE HEAT 6525
## 252 LIGHTNING 5230
## 148 HEAT 2100
## 236 ICE STORM 1975
## 70 FLASH FLOOD 1777
## 365 THUNDERSTORM WIND 1488
## 131 HAIL 1361Tornadoes have the same impact on human population in fatalities as the rest weather events combined and they have twice more impact in injuries.
torn_fat <- c(top10_fat$FATALITIES[1], sum(top10_fat$FATALITIES)-top10_fat$FATALITIES[1])
names(torn_fat) <- c("Tornado", "Others")
torn_inj <- c(top10_inj$INJURIES[1], sum(top10_inj$INJURIES)-top10_inj$INJURIES[1])
names(torn_inj) <- c("Tornado", "Others")
par(mfrow=c(2,2))
barplot(top10_fat$FATALITIES, names=top10_fat$EVTYPE,las=2, col="red",
main="Top 10 most dedliest\n weather events", cex.names=0.6)
barplot(top10_inj$INJURIES, names=top10_inj$EVTYPE,las=2, col="red",
main="Top 10 weather events\n for injuries", cex.names=0.6)
barplot(torn_fat, col=c("red", "black"), main="Tornado vs other weather \n events fatalities")
barplot(torn_inj, col=c("red", "black"), main="Tornado vs other weather \n events injuries")
Q2. Across the United States, which types of events have the greatest economic consequences?
In order to answer this question we need to aggregate property and crops damages by event type. Also we need to calculate the sum of both of them to find the total financial loss.
## Find economical impact of severe weather events
dmg_econ <- aggregate(cbind(PROPDMG_FIN, CROPDMG_FIN, PROPDMG_FIN + CROPDMG_FIN) ~ EVTYPE,
df, FUN=sum)
names(dmg_econ)[4] <- "TOTAL_DMG"
### Find top10 by damages on total, property and crops
top10_total <- head(dmg_econ[order(dmg_econ$TOTAL_DMG, decreasing=T), ], n=10)
top10_prop <- head(dmg_econ[order(dmg_econ$PROPDMG_FIN, decreasing=T), c(1,2)], n=10)
top10_crop <- head(dmg_econ[order(dmg_econ$CROPDMG_FIN, decreasing=T), c(1,3)], n=10)Top 10 weather events that result the biggest financial loss.
top10_total## EVTYPE PROPDMG_FIN CROPDMG_FIN TOTAL_DMG
## 83 FLOOD 144657709807 5661968450 150319678257
## 222 HURRICANE/TYPHOON 69305840000 2607872800 71913712800
## 405 TORNADO 56947380676 414953270 57362333946
## 348 STORM SURGE 43323536000 5000 43323541000
## 131 HAIL 15735267513 3025954473 18761221986
## 70 FLASH FLOOD 16822673978 1421317100 18243991078
## 45 DROUGHT 1046106000 13972566000 15018672000
## 213 HURRICANE 11868319010 2741910000 14610229010
## 308 RIVER FLOOD 5118945500 5029459000 10148404500
## 236 ICE STORM 3944927860 5022113500 8967041360Top 10 weather events that result the biggest property damages.
top10_prop## EVTYPE PROPDMG_FIN
## 83 FLOOD 144657709807
## 222 HURRICANE/TYPHOON 69305840000
## 405 TORNADO 56947380676
## 348 STORM SURGE 43323536000
## 70 FLASH FLOOD 16822673978
## 131 HAIL 15735267513
## 213 HURRICANE 11868319010
## 415 TROPICAL STORM 7703890550
## 481 WINTER STORM 6688497251
## 197 HIGH WIND 5270046295Top 10 weather events that result the biggest crops damages.
top10_crop## EVTYPE CROPDMG_FIN
## 45 DROUGHT 13972566000
## 83 FLOOD 5661968450
## 308 RIVER FLOOD 5029459000
## 236 ICE STORM 5022113500
## 131 HAIL 3025954473
## 213 HURRICANE 2741910000
## 222 HURRICANE/TYPHOON 2607872800
## 70 FLASH FLOOD 1421317100
## 63 EXTREME COLD 1292973000
## 111 FROST/FREEZE 1094086000Floods and droughts result the biggest financial loss. In case of property damages the most dangerous events are floods. In case of crops loss the most dangerous events are droughts. But floods result the biggest total damages.
total_dmg <- c(top10_total$TOTAL_DMG[1], sum(top10_total$TOTAL_DMG)-top10_total$TOTAL_DMG[1])
names(total_dmg) <- c("Flood", "Others")
prop_dmg <- c(top10_prop$PROPDMG_FIN[1], sum(top10_prop$PROPDMG_FIN)-top10_prop$PROPDMG_FIN[1])
names(prop_dmg) <- c("Flood", "Others")
crop_dmg <- c(top10_crop$CROPDMG_FIN[1],
top10_crop$CROPDMG_FIN[2] + top10_crop$CROPDMG_FIN[3],
sum(top10_crop$CROPDMG_FIN)-top10_crop$CROPDMG_FIN[1]-top10_crop$CROPDMG_FIN[2]-
top10_crop$CROPDMG_FIN[3])
names(crop_dmg) <- c("Droughts", "Floods", "Others")
### Plot the results
par(mfrow=c(2,3))
barplot(rbind(top10_total$PROPDMG_FIN, top10_total$CROPDMG_FIN), names=top10_total$EVTYPE, las=2,
col=c("red","green"),
main="Top 10 severe weather events\n by their economical impact", cex.names=0.6)
barplot(top10_prop$PROPDMG_FIN, names=top10_prop$EVTYPE, las=2, col="red",
main="Top10 severe weather events by their \nimpact for damages on property", cex.names=0.6)
barplot(top10_crop$CROPDMG_FIN, names=top10_crop$EVTYPE, las=2, col="green",
main="Top10 severe weather events by their \nimpact for damages on crops", cex.names=0.6)
barplot(total_dmg, col=c("blue", "black"), main="Flood vs other weather \n events by total economical damage")
barplot(prop_dmg, col=c("blue", "black"), main="Flood vs other weather \n events by property damage")
barplot(crop_dmg, col=c("bisque", "blue", "black"), main="Drought and Floods vs other weather \n
events by crops damage", las=2)