Storm Data

# Download the Data set
    fileurl <- "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2"
    download.file(fileurl, destfile = 'repdata_data_StormData.csv.bz2')

# Loading the data set
    rawData <- read.csv("repdata_data_StormData.csv.bz2", header = T)

# Checking the dimension of the data set
    dim(rawData)

# Checking first 5 rows of the data set
    head(rawData, 5)

# Subsetting the data set
    data <- rawData[,c("EVTYPE", "FATALITIES", "INJURIES", "PROPDMG", "PROPDMGEXP", "CROPDMG", "CROPDMGEXP")]

# Assign categories in old and new objects
    old <- c("WIND", "RAIN", "FLOOD", "STROM|STORM", "HAIL|BLIZZARD", "HEAT", "SNOW", "COLD|ICE|FREEZE|FROST", "WINTER", "WILD", "FOG", "LIGHTNING", "DRY|DROUGHT", "TORNADO|HURRICANE")
    new <- c("WIND", "RAIN", "FLOOD", "STORM", "BLIZZARD", "HEAT","SNOW", "FROST", "WINTER", "WILDFIRE", "FOG", "LIGHTNING", "DROUGHT", "TORNADO")

# Replacing EVTYPE with assigned categories
    for (i in 1:14){
        data$EVTYPE[grep(old[i], data$EVTYPE, ignore.case = T)] <- new[i]
    }

# Assign types of measurement in old and new objects
    old <- c("K|M|B", "K", "M", "B")
    new <- c(0, 10^3, 10^6, 10^9)

# Replacing value of PROPDMGEXP and CROPDMGEXP variable
    data$PROPDMGEXP[!grepl(old[1], data$PROPDMGEXP, ignore.case = T)] <- new[1]
    data$CROPDMGEXP[!grepl(old[1], data$CROPDMGEXP, ignore.case = T)] <- new[1]

    for (i in 2:4){
        data$PROPDMGEXP <- gsub(old[i], new[i], data$PROPDMGEXP, ignore.case = T)
        data$CROPDMGEXP <- gsub(old[i], new[i], data$CROPDMGEXP, ignore.case = T)
    }

# Converting PROPDMGEXP and CROPDMGEXP to numeric
    data$PROPDMGEXP <- as.numeric(data$PROPDMGEXP)
    data$CROPDMGEXP <- as.numeric(data$CROPDMGEXP)

# Making a new data set consist of 3 variable, which are EVTYPE, FATALITIES, and INJURIES
    eventHealth <- aggregate(data[c("FATALITIES", "INJURIES")], data["EVTYPE"], sum)

# Arrange the type of event based on INJURIES
    eventHealth <- eventHealth[order(eventHealth$INJURIES, decreasing = T),]

# Subset the top 10 observations in the arranged data set
    eventHealth <- head(eventHealth, 10)

eventHealth

# Reformatting new data set
    # 1. Fatalities
        # Subsetting the data set to only show EVTYPE and FATALITIES variable
            fatalities <- eventHealth[-3] # exclude INJURIES variable
        # Change colnames FATALITIES to total
            colnames(fatalities)[2] <- "Total"
        # Add new column Type with vales "Fatalities" to identify total fatalities
            fatalities$Type <- "Fatalities"

    # 2. Do the same thing to INJURIES variable
        injuries <- eventHealth[-2] # exclude FATALITIES variable
        colnames(injuries)[2] <- "Total"
        injuries$Type <- "Injuries"

    # 3. Merge both FATALITIES and INJURIES data set into one object
        eventHealth <- rbind(fatalities, injuries)

eventHealth

# Making new data set for examine the economic consequences
    eventEconomy <- data[, -(2:3)] # Excluding FATALITIES and INJURIES variable

# Make some variable to examine the Damages variable
    library(dplyr)
    eventEconomy <- eventEconomy %>% select(everything()) %>% mutate(PROPTTLDMG = PROPDMG * PROPDMGEXP,
                                          CROPTTLDMG = CROPDMG * CROPDMGEXP,
                                          TOTALDMG = PROPTTLDMG + CROPTTLDMG) %>% select(EVTYPE, PROPTTLDMG, CROPTTLDMG, TOTALDMG)

# Subset the data set to show ONLY EVTYPE and Total Damage variables
    eventEconomy <- aggregate(eventEconomy[-1], eventEconomy[1], sum)

# Arrange the type of event based on TOTALDMG
    eventEconomy <- eventEconomy[order(eventEconomy$TOTALDMG, decreasing = T),]

# Subset the top 10 observations in the arranged data set
    eventEconomy <- head(eventEconomy, 10)

eventEconomy

# Reformatting new data set
    # 1. Properties
        # Subsetting the data set to only show EVTYPE and PROPTTLDMG variable
            property <- eventEconomy[c(1,2)] 
        # Change colnames PROPTTLDMG to damage
            colnames(property)[2] <- "Damage"
        # Add new column Type with vales "Fatalities" to identify total fatalities
            property$Type <- "Property"

    # 2. Do the same thing to CROPTTLDMG variable
        crop <- eventEconomy[c(1,3)]
        colnames(crop)[2] <- "Damage"
        crop$Type <- "Crop"

    # 3. Merge both FATALITIES and INJURIES data set into one object
        eventEconomy <- rbind(property, crop)

eventEconomy

# Import necessary library
    library(ggplot2)

# Create plot with eventHealth object
    ggplot(eventHealth, aes(x = reorder(EVTYPE, -Total), y = Total, fill = Type)) + 
            facet_wrap(Type~., scales = "free") +
            geom_bar(stat = 'identity') + 
            ggtitle("Health Impact in US based on top 10 Weather Events") + 
            xlab('Type of Weather') + ylab("Number of Health Impact") +
            theme(axis.text.x = element_text(angle = 45, hjust = 1))

# Create plot with eventEconomy object
    ggplot(eventEconomy, aes(x = reorder(EVTYPE, -Damage), y = Damage/10^6, fill = Type)) + 
            facet_wrap(Type~., scales = "free") +
            geom_bar(stat = 'identity') +
            ggtitle("Comparing Property and Crop Total Damages in US\n based on top 10 Weather Events") + 
            xlab("Type of Weather") + 
            ylab(expression("Total Damages per 10"^6* " Dollars")) +
            theme(axis.text.x = element_text(angle = 45, hjust = 1))