Impact of storms and other severe weather events on population health and economy in the US
Dino Budimlija
Synopsis
The objective of this report is to explore NOAA (U.S. National Oceanic & Atmospheric Administration) Storm Database and explore which severe weather events have the most harmful impact on the population health and economy. In particular, we are interested in the counts of estimated population fatalities and injuries, and also the size of property and crop damage linked with the specific weather events. To answer these questions we’ll use the database Storm Data (47MB) from NOAA and provide the analysis code to produce tables, figures and summaries required for answering this questions.
This report is a part of Reproducible Research course in Johns Hopkins University Data Specialization track at Coursera and should be provided as a markdown document and published to RPubs where it will be available for peer assessment.
Data Processing
The dataset for this assignment will be downloaded from National Weather Service database that tracks characteristics of major storms and weather events in the United States, including when and where they occur, as well as estimates of any fatalities, injuries, and property damage. The dataset comes in the form of a comma-separated-value file compressed via the bzip2 algorithm to reduce its size.
Loading the data
First, let’s load the libraries we’ll use for the analysis.
library(dplyr)
library(reshape2)
library(ggplot2)
library(gridExtra) ## Loading required package: gridNext, we’ll download the data from the link provided:
## Create data dir
if(!file.exists("Data")) {
dir.create("Data")
}
URL <- "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2"
## download file if not already downloaded
if (!file.exists("./Data/repdata-data-StormData.csv.bz2")) {
download.file(URL, destfile = "./Data/repdata-data-StormData.csv.bz2")
dateDownloaded <- date()
}Check to see if data has successfully downloaded:
list.files("./Data")## [1] "repdata-data-StormData.csv.bz2"We’ll load the dataset next and check its size:
## load dataset if not already loaded
if (!"raw_data" %in% ls()) {
raw_data <- read.csv(bzfile("./Data/repdata-data-StormData.csv.bz2"))
}
## vars for reporting dataset size
rownum <- dim(raw_data)[1]
colnum <- dim(raw_data)[2]The dataset consists of 37 variables and 902297 observations.
The variables included in the dataset are:
names(raw_data)## [1] "STATE__" "BGN_DATE" "BGN_TIME" "TIME_ZONE" "COUNTY"
## [6] "COUNTYNAME" "STATE" "EVTYPE" "BGN_RANGE" "BGN_AZI"
## [11] "BGN_LOCATI" "END_DATE" "END_TIME" "COUNTY_END" "COUNTYENDN"
## [16] "END_RANGE" "END_AZI" "END_LOCATI" "LENGTH" "WIDTH"
## [21] "F" "MAG" "FATALITIES" "INJURIES" "PROPDMG"
## [26] "PROPDMGEXP" "CROPDMG" "CROPDMGEXP" "WFO" "STATEOFFIC"
## [31] "ZONENAMES" "LATITUDE" "LONGITUDE" "LATITUDE_E" "LONGITUDE_"
## [36] "REMARKS" "REFNUM"There is also some documentation of the database available where we can find how some of the variables are constructed/defined:
National Weather Service Storm Data Documentation
National Climatic Data Center Storm Events FAQ
Codebook for the dataset from National Climatic Data Center
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, so we’ll only use the more recent data to get more consistent result.
Here’s a sample of raw data:
data <- tbl_df(raw_data)
print(data)## Source: local data frame [902,297 x 37]
##
## STATE__ BGN_DATE BGN_TIME TIME_ZONE COUNTY COUNTYNAME STATE
## 1 1 4/18/1950 0:00:00 0130 CST 97 MOBILE AL
## 2 1 4/18/1950 0:00:00 0145 CST 3 BALDWIN AL
## 3 1 2/20/1951 0:00:00 1600 CST 57 FAYETTE AL
## 4 1 6/8/1951 0:00:00 0900 CST 89 MADISON AL
## 5 1 11/15/1951 0:00:00 1500 CST 43 CULLMAN AL
## 6 1 11/15/1951 0:00:00 2000 CST 77 LAUDERDALE AL
## 7 1 11/16/1951 0:00:00 0100 CST 9 BLOUNT AL
## 8 1 1/22/1952 0:00:00 0900 CST 123 TALLAPOOSA AL
## 9 1 2/13/1952 0:00:00 2000 CST 125 TUSCALOOSA AL
## 10 1 2/13/1952 0:00:00 2000 CST 57 FAYETTE AL
## .. ... ... ... ... ... ... ...
## Variables not shown: EVTYPE (fctr), BGN_RANGE (dbl), BGN_AZI (fctr),
## BGN_LOCATI (fctr), END_DATE (fctr), END_TIME (fctr), COUNTY_END (dbl),
## COUNTYENDN (lgl), END_RANGE (dbl), END_AZI (fctr), END_LOCATI (fctr),
## LENGTH (dbl), WIDTH (dbl), F (int), MAG (dbl), FATALITIES (dbl),
## INJURIES (dbl), PROPDMG (dbl), PROPDMGEXP (fctr), CROPDMG (dbl),
## CROPDMGEXP (fctr), WFO (fctr), STATEOFFIC (fctr), ZONENAMES (fctr),
## LATITUDE (dbl), LONGITUDE (dbl), LATITUDE_E (dbl), LONGITUDE_ (dbl),
## REMARKS (fctr), REFNUM (dbl)Omitting the lower quality data from earlier years
We’ll use only the data for the years where number of recorded events is larger than 10,000.
data$Year = as.numeric(format(as.Date(data$BGN_DATE, format = "%m/%d/%Y %H:%M:%S"), "%Y"))
by_years <- group_by(data, Year)
by_years <- summarize(by_years, Event_count = n())
print(by_years)
storm_data <- filter(data, Year >= 1989)## Source: local data frame [62 x 2]
##
## Year Event_count
## 1 1950 223
## 2 1951 269
## 3 1952 272
## 4 1953 492
## 5 1954 609
## 6 1955 1413
## 7 1956 1703
## 8 1957 2184
## 9 1958 2213
## 10 1959 1813
## 11 1960 1945
## 12 1961 2246
## 13 1962 2389
## 14 1963 1968
## 15 1964 2348
## 16 1965 2855
## 17 1966 2388
## 18 1967 2688
## 19 1968 3312
## 20 1969 2926
## 21 1970 3215
## 22 1971 3471
## 23 1972 2168
## 24 1973 4463
## 25 1974 5386
## 26 1975 4975
## 27 1976 3768
## 28 1977 3728
## 29 1978 3657
## 30 1979 4279
## 31 1980 6146
## 32 1981 4517
## 33 1982 7132
## 34 1983 8322
## 35 1984 7335
## 36 1985 7979
## 37 1986 8726
## 38 1987 7367
## 39 1988 7257
## 40 1989 10410
## 41 1990 10946
## 42 1991 12522
## 43 1992 13534
## 44 1993 12607
## 45 1994 20631
## 46 1995 27970
## 47 1996 32270
## 48 1997 28680
## 49 1998 38128
## 50 1999 31289
## 51 2000 34471
## 52 2001 34962
## 53 2002 36293
## 54 2003 39752
## 55 2004 39363
## 56 2005 39184
## 57 2006 44034
## 58 2007 43289
## 59 2008 55663
## 60 2009 45817
## 61 2010 48161
## 62 2011 62174From this summary we find that for years 1989-2011 we have sufficient data to satisfy our requirement, so we’ll use only that period data for our analysis.
The main variables of our interest here are:
- EVTYPE - Event Type
- MAG - Magnitude
- FATALITIES - Number of fatalities
- INJURIES - Number of persons injured
- PROPDMG - Property damage [USD]
- PROPDMGEXP - Property damage exponent (thousands, millions, billions…)
- CROPDMG - Crops damage [USD]
- CROPDMGEXP - Crops damage exponent (thousands, millions, billions…)
We’ll clean up the data and produce the subset data table with only those variables we are interested in.
storm_data <- subset(storm_data, select=c(8, 23:28))
names(storm_data) <- c("Event_Type", "Fatalities", "Injuries", "Property_Damage", "Property_Damage_Exp", "Crop_Damage", "Crop_Damage_Exp")
str(storm_data)## Classes 'tbl_df' and 'data.frame': 762150 obs. of 7 variables:
## $ Event_Type : Factor w/ 985 levels " HIGH SURF ADVISORY",..: 834 834 856 834 856 856 856 244 856 856 ...
## $ Fatalities : num 0 0 0 0 0 0 0 0 0 0 ...
## $ Injuries : num 0 0 0 0 0 0 0 0 0 0 ...
## $ Property_Damage : num 2.5 250 0 2.5 0 0 0 0 0 0 ...
## $ Property_Damage_Exp: Factor w/ 19 levels "","-","?","+",..: 19 17 1 19 1 1 1 1 1 1 ...
## $ Crop_Damage : num 0 0 0 0 0 0 0 0 0 0 ...
## $ Crop_Damage_Exp : Factor w/ 9 levels "","?","0","2",..: 1 1 1 1 1 1 1 1 1 1 ...PROPDMGEXP and CROPDMGEXP variables denote the multiplier by which we need to multiply corresponding PROPDMG and CROPDMG columns. So the total property and crops damage is calculated like this:
- Property_Damage = PROPDMG * 10 ^ PROPDMGEXP
- Crop_Damage = CROPDMG * 10 ^ CROPDMGEXP
Now, we’ll use those 4 original columns create one column for total Property damage and one for total Crop damage.
# Check the levels for exponent variables
unique(storm_data$Property_Damage_Exp)
unique(storm_data$Crop_Damage_Exp)## [1] M K B m + 0 5 6 ? 4 2 3 h 7 H - 1 8
## Levels: - ? + 0 1 2 3 4 5 6 7 8 B h H K m M
## [1] M K m B ? 0 k 2
## Levels: ? 0 2 B k K m MWe can see that level values for Property and Crop damage exponents are not uniform and are represented as numbers, characters (k, K, m, M, b, B) or even special characters (-, +, ?) and some are simply missing. We’ll convert those level values to meaningful integers for the exponents so we can use them in making single column for Property and Crop damage values (in dollars). All the characters we can’t interpret or are missing we’ll substitute with zeroes.
# Make list of all different exponent values as integers
exp <- list("0" = c("-", "+", "?", "0", "", " "), "2" = c("2", "h", "H"), "3" = c("3", "k", "K"), "6" = c("6", "m", "M"), "9" = c("9", "b", "B"))
# Rename levels for exponent data
levels(storm_data$Property_Damage_Exp) <- exp
levels(storm_data$Crop_Damage_Exp) <- exp
# Convert exponent variables from Factor to Integer
storm_data$Property_Damage_Exp <- as.numeric(as.character(storm_data$Property_Damage_Exp))
storm_data$Crop_Damage_Exp <- as.numeric(as.character(storm_data$Crop_Damage_Exp))
# Create new variables
storm_data <- mutate(storm_data, Property_Damage = Property_Damage * 10^Property_Damage_Exp)
storm_data <- mutate(storm_data, Crop_Damage = Crop_Damage * 10^Crop_Damage_Exp)
# Remove redundant exponent columns
storm_data <- select(storm_data, -c(5, 7))The prepared dataset we’ll work with now looks like this after tidying up:
print(storm_data, )## Source: local data frame [762,150 x 5]
##
## Event_Type Fatalities Injuries Property_Damage Crop_Damage
## 1 TORNADO 0 0 2500000 0
## 2 TORNADO 0 0 250000 0
## 3 TSTM WIND 0 0 0 0
## 4 TORNADO 0 0 2500000 0
## 5 TSTM WIND 0 0 0 0
## 6 TSTM WIND 0 0 0 0
## 7 TSTM WIND 0 0 0 0
## 8 HAIL 0 0 0 0
## 9 TSTM WIND 0 0 0 0
## 10 TSTM WIND 0 0 0 0
## .. ... ... ... ... ...Results
The data analysis must address the following questions:
Across the United States, which types of events (as indicated in the EVTYPE variable) are most harmful with respect to population health?
Across the United States, which types of events have the greatest economic consequences?
The most deadly/harmful weather events in the US from 1989-2011
We’ll group the dataset by the event type and make summary to find out which weather events are the most harmful for human lives.
by_event <- group_by(storm_data, Event_Type)
summary_inj_fat <- summarize(by_event, sum(Injuries), sum(Fatalities))
names(summary_inj_fat)[2:3] <- c("Injuries", "Fatalities")
summary_inj_fat <- mutate(summary_inj_fat, Total_injuries_fatalities = Injuries + Fatalities)
top_inj_fat <- arrange(summary_inj_fat, desc(Total_injuries_fatalities))
head(top_inj_fat, n = 15)## Source: local data frame [15 x 4]
##
## Event_Type Injuries Fatalities Total_injuries_fatalities
## 1 TORNADO 27944 1802 29746
## 2 EXCESSIVE HEAT 6525 1903 8428
## 3 FLOOD 6789 470 7259
## 4 LIGHTNING 5230 816 6046
## 5 TSTM WIND 5404 356 5760
## 6 HEAT 2100 937 3037
## 7 FLASH FLOOD 1777 978 2755
## 8 ICE STORM 1975 89 2064
## 9 THUNDERSTORM WIND 1488 133 1621
## 10 WINTER STORM 1321 206 1527
## 11 HIGH WIND 1137 248 1385
## 12 HURRICANE/TYPHOON 1275 64 1339
## 13 HAIL 1162 15 1177
## 14 HEAVY SNOW 1021 127 1148
## 15 WILDFIRE 911 75 986Since we can see from this summary that we have some inconsistent naming for the Event_Type variable values, we need to clean up those values and merge some of this event names into one umbrella category. This list of values is compiled using more than 100 most dangerous weather event categories for human lives and material property before tidying them up. After cleaning up we’ll have 16 weather event categories we’ll use in analysis.
event_recode <- list(
"Cold Weather/Snow" = c("ICE STORM", "WINTER STORM", "AVALANCHE", "HEAVY SNOW", "BLIZZARD", "WINTER WEATHER/MIX", "WINTRY MIX", "WINTER WEATHER MIX", "SNOW SQUALL", "SNOW/HIGH WINDS", "SNOW", "EXTREME WINDCHILL", "WINTER STORM HIGH WINDS", "BLOWING SNOW", "COLD AND SNOW", "WINTER WEATHER", "EXTREME COLD", "EXTREME COLD/WIND CHILL", "COLD/WIND CHILL", "COLD", "HEAVY SNOW/ICE", "HIGH WINDS/SNOW", "WINTER STORM"),
"Fire" = c("WILDFIRE", "WILD/FOREST FIRE", "WILD FIRES", "HIGH WINDS/COLD"),
"Flood" = c("FLOOD", "FLASH FLOOD", "URBAN/SML STREAM FLD", "FLOOD/FLASH FLOOD", "FLASH FLOOD/FLOOD", "WATERSPOUT", "FLASH FLOODING", "FLASH FLOOD/FLOOD", "FLOODING", "RIVER FLOOD", "COASTAL FLOOD", "River Flooding", "COASTAL FLOODING", "FLOOD/RAIN/WINDS", "MAJOR FLOOD"),
"Fog" = c("FOG", "DENSE FOG"),
"Hail" = c("HAIL", "TSTM WIND/HAIL", "SMALL HAIL", "HAILSTORM"),
"Heat" = c("EXCESSIVE HEAT", "HEAT", "HEAT WAVE", "EXTREME HEAT", "Heat Wave", "RECORD HEAT", "DRY MICROBURST", "UNSEASONABLY WARM AND DRY", "UNSEASONABLY WARM", "HEAT WAVE DROUGHT", "RECORD/EXCESSIVE HEAT", "DROUGHT"),
"Hurricane" = c("HURRICANE/TYPHOON", "HURRICANE", "HURRICANE OPAL", "HURRICANE ERIN", "HURRICANE OPAL/HIGH WINDS"),
"Ice" = c("GLAZE", "ICE", "ICY ROADS", "FREEZING RAIN", "FREEZING DRIZZLE", "GLAZE/ICE STORM", "ICE STORM", "FROST/FREEZE", "FREEZE", "DAMAGING FREEZE", "FROST"),
"Land Slide" = c("LANDSLIDE"),
"Rain" = c("HEAVY RAIN", "HEAVY RAIN/SEVERE WEATHER, RAIN", "MIXED PRECIP", "EXCESSIVE RAINFALL", "EXCESSIVE WETNESS", "HEAVY RAINS"),
"Rip Current" = c("RIP CURRENT", "RIP CURRENTS"),
"Thunderstorm" = c("TSTM WIND", "LIGHTNING", "STORM SURGE", "MARINE THUNDERSTORM WIND", "THUNDERSTORMW", "SEVERE THUNDERSTORM", "THUNDERSTORM WINDS", "MARINE TSTM WIND", "THUNDERSTORM", "THUNDERSTORM WINDS", "THUNDERSTORM WIND"),
"Tornado" = c("TORNADO", "WATERSPOUT/TORNADO", "TORNADOES, TSTM WIND, HAIL", "TORNADO F2"),
"Tropical Storm" = c("TROPICAL STORM", "DUST STORM", "TROPICAL STORM GORDON", "DUST DEVIL"),
"Tsunami/High Surf" = c("HIGH SURF", "TSUNAMI", "HEAVY SURF/HIGH SURF", "HEAVY SURF", "HIGH WIND AND SEAS", "STORM SURGE/TIDE", "HIGH SEAS", "ROUGH SEAS", "MARINE MISHAP", "High Surf", "HIGH SURF ADVISORY"),
"Wind" = c("THUNDERSTORM WIND", "HIGH WIND", "STRONG WIND", "HIGH WINDS", "WIND", "MARINE STRONG WIND", "STRONG WINDS", "GUSTY WINDS"))
levels(top_inj_fat$Event_Type) <- event_recode
top_inj_fat <- group_by(top_inj_fat, Event_Type)
top_inj_fat <- summarize(top_inj_fat, sum(Injuries), sum(Fatalities), sum(Total_injuries_fatalities))
names(top_inj_fat)[2:4] <- c("Injuries", "Fatalities", "Total_casualties")
top_inj_fat <- arrange(top_inj_fat, desc(Total_casualties))
top_inj_fat[1:15, ]## Source: local data frame [15 x 4]
##
## Event_Type Injuries Fatalities Total_casualties
## 1 Tornado 28002 1830 29832
## 2 Thunderstorm 11663 1269 12932
## 3 Heat 9273 3174 12447
## 4 Flood 8704 1541 10245
## 5 Cold Weather/Snow 4396 1178 5574
## 6 Wind 3344 567 3911
## 7 Ice 2415 118 2533
## 8 Fire 1610 90 1700
## 9 Hurricane 1323 134 1457
## 10 Hail 1267 20 1287
## 11 Fog 1076 80 1156
## 12 Rip Current 529 572 1101
## 13 Tropical Storm 865 90 955
## 14 Tsunami/High Surf 416 220 636
## 15 Rain 302 102 404Let’s plot the top 10 weather events with highest casualties count in the US from 1989-2011.
# Melting the data for plotting
top_inj_fat <- transform(top_inj_fat, Event_Type = reorder(Event_Type, order(Total_casualties, decreasing = TRUE))) # reorder Event_Type factor levels from more casualties to less for proper plotting later on
top_inj_fat_melted <- melt(top_inj_fat[1:10 ,1:3], id = "Event_Type") # melt top-10 events
names(top_inj_fat_melted)[2:3] <- c("Casualty_Type", "Count")
ggplot(data = top_inj_fat_melted, aes(x = Event_Type, y = Count, fill = Casualty_Type)) +
geom_bar(stat = "identity") +
labs(title = "Most harmful weather events in the US from 1989-2011", x = "Weather event", y = "Number of persons injured/killed") +
scale_fill_manual(values=c("blue", "red")) +
theme(axis.text.x = element_text(angle = 30, hjust = 1, vjust = 1, face="bold"))
We can see that tornado is holding the first position as the most dangerous weather event causing the human injuries and fatalities. We can also observe that thunderstorms, heat, floods, cold weather/snows and wind are also very dangerous when looking at injuries inflicted, as well as fatalities, but we can see that the order regarding the fatalities is somewhat different with **heat* causing the most fatalities.
We can consider other weather events little less dangerous since their casualty count is lower, with very small count of fatalities.
To make better comparation between weather events causing the most injuries and the ones resulting in the most fatalities, we’ll make side by side plot.
top_fatalities <- top_inj_fat[, c(1,3)]
top_injuries <- top_inj_fat[, c(1,2)]
top_fatalities <- arrange(top_fatalities, desc(Fatalities))
top_injuries <- arrange(top_injuries, desc(Injuries))
top_fatalities <- transform(top_fatalities, Event_Type = reorder(Event_Type, order(Fatalities, decreasing = TRUE)))
top_injuries <- transform(top_injuries, Event_Type = reorder(Event_Type, order(Injuries, decreasing = TRUE)))
Inj_plot <- ggplot(top_injuries[1:8, ], aes(x = Event_Type, y = Injuries)) +
geom_bar(stat = "identity", fill = "blue", color = "black") +
labs(x = "Weather event type", y = "Number of injuries") +
coord_cartesian(ylim = c(0, 30000), xlim=c(0, 9)) +
theme(axis.text.x = element_text(angle = 30, hjust = 1, vjust = 1))
Fat_plot <- ggplot(top_fatalities[1:8, ], aes(x = Event_Type, y = Fatalities)) +
geom_bar(stat = "identity", fill = "red", color = "black") +
labs(x = "Weather event type", y = "Number of fatalities") +
coord_cartesian(ylim = c(0, 30000), xlim=c(0, 9)) +
theme(axis.text.x = element_text(angle = 30, hjust = 1, vjust = 1))
grid.arrange(arrangeGrob(Inj_plot, Fat_plot, ncol = 2, main = textGrob("Most harmful weather events causing injuries vs fatalities", vjust = 0.5, gp = gpar(fontface = "bold"))))
From this plot we can conclude that heat really is the most fatal weather event, followed by tornadoes, floods, thunderstorms and cold weather/snow.
Weather events causing the highest damage to properties and crops in the US from 1989-2011
We’ll summarize the data to find out what types of severe weather events inflict the greatest damage to material property in the US.
# Clean up the datase, use new categories for weather events, excluse data with NAs for Property_Damage
levels(by_event$Event_Type) <- event_recode
by_event <- data.frame(by_event)
damage <- tbl_df(by_event[, c(1, 4, 5)])
Property_NAs <- complete.cases(damage$Property_Damage)
damage <- damage[which(Property_NAs == TRUE), ] # remove Property_Damage NA's
# Make all unspecified weather events as "Other" category
levels(damage$Event_Type)[which(is.na(damage$Event_Type))] <- "Other"
damage$Event_Type[which(is.na(damage$Event_Type))] <- "Other"
damage$Event_Type <- factor(damage$Event_Type)
# Summarize the data with greatest economic consequences
damage <- group_by(damage, Event_Type)
summary_damage <- summarize(damage, sum(Property_Damage), sum(Crop_Damage))
names(summary_damage)[2:3] <- c("Property_Damage", "Crop_Damage")
summary_damage <- mutate(summary_damage, Total_damage = Property_Damage + Crop_Damage)
summary_damage <- arrange(summary_damage, desc(Total_damage))
summary_damage <- transform(summary_damage, Event_Type = reorder(Event_Type, order(Total_damage, decreasing = TRUE))) # reorder Event_Type factor levels from more total damage to less for proper plotting later on
head(summary_damage, n = 15)
summary_damage <- summary_damage[, 1:3]## Event_Type Property_Damage Crop_Damage Total_damage
## 1 Flood 167423099313 12381239200 179804338513
## 2 Hurricane 84705105010 5514792800 90219897810
## 3 Thunderstorm 51750392734 758009228 52508401962
## 4 Tornado 33871276557 417453270 34288729827
## 5 Hail 16017673013 3086443723 19104116736
## 6 Heat 1073164350 14877054500 15950218850
## 7 Ice 3984720560 6890524500 10875245060
## 8 Wind 9548504585 1164588450 10713093035
## 9 Cold Weather/Snow 8467308091 1604290100 10071598191
## 10 Fire 8501543500 409269630 8910813130
## 11 Tropical Storm 7710639880 681946000 8392585880
## 12 Tsunami/High Surf 4886530500 870000 4887400500
## 13 Other 3670999916 336363980 4007363896
## 14 Rain 706533140 935899800 1642432940
## 15 Land Slide 324596000 20017000 344613000We can see the top 15 weather events with the most negative impact on the economy in the above summary.
# Melting the data for plotting
summary_damage_melted <- melt(summary_damage[1:10, ], id = "Event_Type") # melt top-10 events
names(summary_damage_melted)[2:3] <- c("Damage_Type", "Value")
summary_damage_melted$Value <- round(summary_damage_melted$Value/1000000, 0)
ggplot(data = summary_damage_melted, aes(x = Event_Type, y = Value, fill = Damage_Type)) +
geom_bar(stat = "identity") +
labs(title = "Weather events that caused most damage in the US from 1989-2011", x = "Weather event", y = "Millions of Dollars") +
scale_fill_manual(values=c("blue", "green")) +
theme(axis.text.x = element_text(angle = 30, hjust = 1, vjust = 1, face="bold"))
From this plot we can conclude that floods are by far the most costly for the economy, followed by hurricanes, thunderstorms and tornadoes with the highest property damage inflicted . It’s important to notice here that hurricane is not the same event as tornado; yearly in the US there are 10-15 hurricanes and about 1200 tornadoes - read more about difference between hurricanes and tornadoes here.
Regarding the damage to crops, we observe the heat as having the most negative effect, followed by the floods, ice and hurricanes.
This document was produced with RStudio v0.98.1091 on R v3.1.2.