Reproducible Research: Peer Assessment 2 - Data Analysis for Severe Weather in US

Assignment Introduction

The goal of this assignment is to explore the NOAA Storm Database and provide a short report about severe weather events in US in order to inform the government or the municipal manager who might be responsible for preparing for severe weather events and will need to prioritize resources for different types of events. The questions to be answered in this report are as follows:

  1. Across the United States, which types of events (as indicated in the EVTYPE variable) are most harmful with respect to population health?
  2. Across the United States, which types of events have the greatest economic consequences?

some extra requirement to be noted

  1. If preprocessing is time-consuming you may consider using the cache = TRUE option for certain code chunks.
  2. The analysis document must have at least one figure containing a plot, but no more than 3 figures (panel plots are allowed).
  3. Figures should have descriptive captions.
  4. Data transformation should be described.

Data Processing

Download the dataset from the course website and unzipt it to get the csv file.

if (!file.exists("data/repdata-data-StormData.csv")) {
    # check if the csv file is available
    dir.create("data", showWarnings = FALSE)  # in case the directory 'data' hasn't been created
    download.file(url = "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2", 
        destfile = "data/repdata-data-StormData.csv.bz2")
    if (!require(R.utils)) {
        install.packages("R.utils")
        library(R.utils)  # a library for unzipping bz2 files
    }
    bunzip2(filename = "data/repdata-data-StormData.csv.bz2", destname = "data/repdata-data-StormData.csv")  # get the csv file for the task
}
df <- read.csv("data/repdata-data-StormData.csv")

Across the United States, which types of events (as indicated in the EVTYPE variable) are most harmful with respect to population health?

In the dataset, there are two columns related to population health, i.e. FATALITIES and INJURIES. The required information is the total number of these two columns for each type of events.

First, we sum up these two columns and put the results into a new column called 'health'.

df$health <- df$FATALITIES + df$INJURIES
# df <- df[,c('EVTYPE','health')]

I categorize all the types of weather events into 6 kinds: cold, heat, water, wind, lightning, and earth. 


cold <- c("avalanche", "avalance", "blizzard", "chill", "cold", "cool", "glaze", 
    "hypothermia", "hyperthermia", "ice", "icy", "freez", "frost", "low temp", 
    "sleet", "snow", "wint")
cold <- cbind(cold, code = rep("cold", length(cold)))
heat <- c("below normal precip", "dry", "drie", "drought", "fire", "heat", "high temp", 
    "hot", "warm")
heat <- cbind(heat, code = rep("heat", length(heat)))
water <- c("coast", "cstl", "current", "dam fail", "dam break", "drizzle", "drown", 
    "erosion", "erosin", "flood", "floood", "fld", "heavy shower", "high water", 
    "high waves", "lake", "landslump", "marine", "precip", "rain", "rising water", 
    "river", "rogue wave", "slide", "stream", "sea", "seiche", "surf", "swell", 
    "tide", "tidal", "torrent", "wet", "fog", "vog")
water <- cbind(water, code = rep("water", length(water)))
wind <- c("burst", "cloud", "depression", "floyd", "funnel", "gust", "hail", 
    "hurricane", "landspout", "storm", "southeast", "thunderstorm", "thundertsorm", 
    "thundestorm", "tornado", "torndao", "tstm", "turbulence", "typhoon", "wall", 
    "waterspout", "water spout", "wayterspout", "wind", "wnd")
wind <- cbind(wind, code = rep("wind", length(wind)))
lightning <- c("lightning", "ligntning", "lighting")
lightning <- cbind(lightning, code = rep("lightning", length(lightning)))
earth <- c("tsunami", "volcan", "dust")
earth <- cbind(earth, code = rep("earth", length(earth)))

events <- data.frame(rbind(cold, heat, water, wind, lightning, earth))
names(events) <- c("EVTYPE", "code")
events$code <- as.character(events$code)

Make the EVTYPE column lower cases for the convenience of matching keywords. 

df$EVTYPE <- tolower(df$EVTYPE)
df$code <- rep("other", nrow(df))

Creat a new column 'string' filled with EVTYPE in regular expression.Then use grep function to get the indeces of the matched rows. Assign the 'code' to their 'code' column of the matched rows. Note that 'lack of snow' was mistaken as 'snow' in this for loop, so I look up this EVTYPE and assign its code as 'other'.

events$string <- paste(".*", events[, "EVTYPE"], ".*", sep = "")
for (i in seq(nrow(events))) {
    df[grep(events$string[i], df$EVTYPE, ignore.case = TRUE), "code"] <- events$code[i]
}
df[grep(".*lack of snow.*", df$EVTYPE, ignore.case = TRUE), "code"] <- "other"
# df$code <- factor(df$code)

Have an overview of the newly added column 'code'.

table(df$code)

## 
##      cold     earth      heat lightning     other     water      wind 
##     30892       638      9615     15780       240    104126    741006

Next, we calculate the overall sum of population health for each type of categories (i.e. cold, heat, water, wind, lightning, and earth), and re-arrange the dataset in a descending order of the number of population health.

healthData <- aggregate(. ~ code, data = df[, c("health", "code")], FUN = sum)
healthData <- healthData[order(healthData$health, decreasing = TRUE), ]

Plot the general kinds of events that are harmful to population health.

barplot(healthData$health, names.arg = healthData$code)
title(xlab = "General Types of Events", ylab = "Sum of Fatalities and Injuries", 
    main = "The Damage to Population Health Caused by Severe Weather in the US")

plot of chunk unnamed-chunk-6

Report the most harmful event

allWindEvents <- paste(tolower(events[events$code == "wind", "EVTYPE"]), collapse = ", ")
paste("The most harmful kinds of events is ", healthData[1, "code"], ", having influenced the health of ", 
    healthData[1, "health"], " people, including fatalities and injuries. This kind of events includes ", 
    allWindEvents, ".")

## [1] "The most harmful kinds of events is  wind , having influenced the health of  117107  people, including fatalities and injuries. This kind of events includes  burst, cloud, depression, floyd, funnel, gust, hail, hurricane, landspout, storm, southeast, thunderstorm, thundertsorm, thundestorm, tornado, torndao, tstm, turbulence, typhoon, wall, waterspout, water spout, wayterspout, wind, wnd ."

Across the United States, which types of events have the greatest economic consequences?

Take a look at the overview of the column PROPDMGEXP

table(df$PROPDMGEXP)

## 
##             -      ?      +      0      1      2      3      4      5 
## 465934      1      8      5    216     25     13      4      4     28 
##      6      7      8      B      h      H      K      m      M 
##      4      5      1     40      1      6 424665      7  11330

I decided to calculate to million, so I create a new column 'million'. I put m & M as one billion, and B as 1000 million, and ignore the rest.

df$million <- rep(0, nrow(df))
df[df$PROPDMGEXP == "B", "million"] <- df[df$PROPDMGEXP == "B", "million"] + 
    1000
df[df$PROPDMGEXP == "M", "million"] <- df[df$PROPDMGEXP == "M", "million"] + 
    1
df[df$PROPDMGEXP == "m", "million"] <- df[df$PROPDMGEXP == "m", "million"] + 
    1

Next, we calculate the overall sum of the economic consequences for each type of categories (i.e. cold, heat, water, wind, lightning, and earth), and re-arrange the dataset in a descending order of the number of economic consequences.

economicData <- aggregate(. ~ code, data = df[, c("million", "code")], FUN = sum)
economicData <- economicData[order(economicData$million, decreasing = TRUE), 
    ]

Plot the economic consequences caused by the general kinds of events .

barplot(economicData$million, names.arg = economicData$code)
title(xlab = "General Types of Events", ylab = "Sum of Economic Consequences (in million)", 
    main = "The Economic Consequences Caused by Severe Weather in the US")

plot of chunk unnamed-chunk-11

Report the event with the greatest economic consequences

paste("The event that causes greatest economic consequences is ", economicData[1, 
    "code"], ", having caused ", economicData[1, "million"], " million of economic consequences. This kind of events includes ", 
    allWindEvents, ".")

## [1] "The event that causes greatest economic consequences is  wind , having caused  37379  million of economic consequences. This kind of events includes  burst, cloud, depression, floyd, funnel, gust, hail, hurricane, landspout, storm, southeast, thunderstorm, thundertsorm, thundestorm, tornado, torndao, tstm, turbulence, typhoon, wall, waterspout, water spout, wayterspout, wind, wnd ."

Results

Accoridng to the above analysis, the event across the US that harms population health most is related to 'wind'.

As for the impacts on economy, the event of 'wind' causes the greatest consequences.