This project for the Coursera data products course uses NOAA data to map out deaths, injuries and dollar damage across the US from weather events.
# reading the data
dataRaw <- read.csv("stormdata_2013.csv")
# filling empty damage values with 0's
dataRaw$DAMAGE_CROPS[dataRaw$DAMAGE_CROPS == ""] <- "0.00K"
dataRaw$DAMAGE_PROPERTY[dataRaw$DAMAGE_PROPERTY == ""] <- "0.00K"
# helper function to expand abbreviated damage dollar value
dollarNum <- function(KM) {
modC <- substr(KM, nchar(KM), nchar(KM))
modN <- case_when(
modC == "K" ~ 1e3,
modC == "M" ~ 1e6,
modC == "B" ~ 1e9,
.default = 1
)
num <- as.numeric(substr(KM, 1, nchar(KM)-1))
num * modN
}
# summarizing impacts by state
dataSum <- dataRaw %>%
group_by(STATE) %>%
summarize(
events = n(),
injuries = sum(INJURIES_DIRECT) + sum(INJURIES_INDIRECT),
deaths = sum(DEATHS_DIRECT) + sum(DEATHS_INDIRECT),
damage = sum(dollarNum(DAMAGE_PROPERTY)) + sum(dollarNum(DAMAGE_CROPS))
)
# converting full uppercase state names to abbreviation used by plotly choropleth
dataSum$STATE <- tools::toTitleCase(tolower(dataSum$STATE))
dataSum$stateAbb <- state.abb[match(dataSum$STATE, state.name)]
dataSum$hover <- with(dataSum, paste(STATE, ":", events, "weather events <br>",
injuries, "injuries,", deaths, "deaths <br> $",
damage, "in damage to property and crops"))
dataSum <- dataSum %>%
select(stateAbb, events, hover) %>%
filter(!is.na(stateAbb))
borders <- list(color = toRGB("black"))
map_options <- list(
scope = 'usa',
projection = list(type = 'albers usa'),
showlakes = T
)
dataSum <- as.data.frame(dataSum)
plot_ly(dataSum, z = dataSum$events, text = dataSum$hover, locations = dataSum$stateAbb,
type = 'choropleth', locationmode = 'USA-states',
color = dataSum$events, colors = 'Reds', marker = list(line = borders)) %>%
layout(title = 'US weather events impacts for 2013', geo = map_options)
Cleanup
rm(list=ls())