Storms and other severe weather events can cause both public health and economic problems for communities and municipalities. Many severe events can result in fatalities, injuries, and property damage, and preventing such outcomes to the extent possible is a key concern.
This project involves exploring the U.S. National Oceanic and Atmospheric Administration’s (NOAA) storm database. This database 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.
Data analysis must address the following questions:
As a first step, I automated the process of download and save the data in a directory /data (also created by the Rmarkdown).
filename <- "./data/2FStormData.csv.bz2"
datapath <- "./data"
# Checking if dir exists
if (!file.exists(datapath)) { dir.create(datapath) }
# Checking if archieve already exists.
if (!file.exists(filename)){
fileUrl <- "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2"
download.file(fileUrl, "./data/2FStormData.csv.bz2")
}
# Reading data
data <- read_csv("./data/2FStormData.csv.bz2",
col_types = cols(EVTYPE = col_character(),
STATE = col_character(),
FATALITIES = col_number(),
INJURIES = col_number(),
PROPDMG = col_number(),
PROPDMGEXP = col_character(),
CROPDMG = col_number(),
CROPDMGEXP = col_character() ))We are interested in variables related with health and economic damage, so I picked the the group of variables:
Health variables:
Economic variables:
Event variable (target variable):
Extras
With these variables I made 4 datasets:
tidydata: (902297 rows, 8 variables). Dataset with all the variables.
harmful_events: (977 rows, 3 variables). Event, Deads, Injuries.
damageData_prop: (436035 rows, 4 variables). EVTYPE (event), PROPDMG, PROPDMGEXP. Added an extra variable millions to unify Economic losses scale.
damageData_crop: (283835 rows, 4 variables). EVTYPE (event), CROPDMG, CROPDMGEXP, Added an extra variable millions to unify Economic losses scale.
# Selecting just useful variables
tidydata <- select(data, c("EVTYPE", "STATE", "FATALITIES", "INJURIES", "PROPDMG",
"PROPDMGEXP", "CROPDMG", "CROPDMGEXP"))
# Making dataset to use in health impact analysis
harmful_events <- tidydata %>% group_by(EVTYPE) %>%
summarise(across(c(FATALITIES,INJURIES),sum)) %>%
rename( Event = EVTYPE, Deads = FATALITIES, Injuries = INJURIES)## `summarise()` ungrouping output (override with `.groups` argument)more_deads <- harmful_events %>% arrange(desc(Deads)) %>% .[1:15,]
more_injuries <- harmful_events %>% arrange(desc(Injuries)) %>% .[1:15,]
# Making a dataset to use in economic analisys
damageData_prop <- filter(tidydata, tidydata$PROPDMGEXP == "K" |
tidydata$PROPDMGEXP == "M" |
tidydata$PROPDMGEXP == "B" ) %>%
select(c(EVTYPE,PROPDMG,PROPDMGEXP))
damageData_crop <- filter(tidydata, tidydata$CROPDMGEXP == "K" |
tidydata$CROPDMGEXP == "M" |
tidydata$CROPDMGEXP == "B" ) %>%
select(c(EVTYPE,CROPDMG,CROPDMGEXP)) Making plot Top 15. Most Frequent Events Across the Country
events <- data.frame(table(tidydata$EVTYPE)) %>% rename( event = Var1 )
most_freq_events <- filter(events, events$Freq >= 100) %>% arrange(desc(Freq))
g_freq <- most_freq_events[1:15,] %>% mutate(event = reorder(event, Freq)) %>%
ggplot(aes(Freq, event)) +
geom_col(position = "dodge") +
geom_text(aes(label = Freq), position = position_dodge(0.9), hjust = -0.1 ) +
xlab("") + ylab("") +
coord_cartesian(xlim = c(0,310000)) +
ggtitle("Top 15. Most Frequent Events Across the Country")Making plot Top 15. Number Of fatalities (A) / Injuries (B)
g_deads <- more_deads %>% mutate(Event = reorder(Event, Deads)) %>%
ggplot(aes(Deads, Event)) +
geom_col(position = "dodge") +
geom_text(aes(label = Deads),size =2.5,
position = position_dodge(0.9), hjust = -0.1 ) +
xlab("") + ylab("") +
coord_cartesian(xlim = c(0,6500)) +
ggtitle("A")
g_injuries <- more_injuries %>% mutate(Event = reorder(Event, Injuries)) %>%
ggplot(aes(Injuries, Event)) +
geom_col(position = "dodge") +
geom_text(aes(label = Injuries), size =2.5,
position = position_dodge(0.9), hjust = -0.1 ) +
xlab("") + ylab("") +
coord_cartesian(xlim = c(0,110000)) +
ggtitle("B")Making Top 15. Economic Losses Properties (A) / Crops (B)
damageData_prop <- damageData_prop %>%
mutate(millions = ifelse(PROPDMGEXP == "K" | PROPDMGEXP == "B",
ifelse(PROPDMGEXP == "K",
PROPDMG/(10^3),PROPDMG*(10^3)), PROPDMG))
damageData_crop <- damageData_crop %>%
mutate(millions = ifelse(CROPDMGEXP == "K" | CROPDMGEXP == "B",
ifelse(CROPDMGEXP == "K",
CROPDMG/(10^3),CROPDMG*(10^3)), CROPDMG))
g_prop <- damageData_prop %>% group_by(EVTYPE) %>%
summarise(across(millions, ~ sum(.x, na.rm = TRUE))) %>%
arrange(desc(millions)) %>% .[1:15,] %>%
mutate(Event = reorder(EVTYPE, millions)) %>%
ggplot(aes(millions, Event)) +
geom_col(position = "dodge") +
geom_text(aes(label = round(millions)),size =2.5,
position = position_dodge(0.9), hjust = -0.1 ) +
xlab("Millions") + ylab("") +
coord_cartesian(xlim = c(0,180000)) +
ggtitle("A")## `summarise()` ungrouping output (override with `.groups` argument)g_crop <- damageData_crop %>% group_by(EVTYPE) %>%
summarise(across(millions, ~ sum(.x, na.rm = TRUE))) %>%
arrange(desc(millions)) %>% .[1:15,] %>%
mutate(Event = reorder(EVTYPE, millions)) %>%
ggplot(aes(millions, Event)) +
geom_col(position = "dodge") +
geom_text(aes(label = round(millions)),size =2.5,
position = position_dodge(0.9), hjust = -0.1 ) +
xlab("Millions") + ylab("") +
coord_cartesian(xlim = c(0,16500)) +
ggtitle("B")## `summarise()` ungrouping output (override with `.groups` argument)g_freq
We can see that tornados are the most harmful event both in injuries and fatalities related.We can see that tornadoes are the most harmful event both in injuries and fatalities caused.
gridExtra::grid.arrange(g_deads,g_injuries,ncol = 2,
top = "Top 15. Number Of fatalities (A) / Injuries (B)")
It seems that weather events are more harmful for properties than for crops. And floods are the cause of the highest economic losses.
gridExtra::grid.arrange(g_prop,g_crop,ncol = 2,
top = "Top 15. Economic Losses Properties (A) / Crops (B)")