Synopsis

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 data analysis involves exploring the U.S. National Oceanic and Atmospheric Administration’s (NOAA) storm database from 1950 to 2011. 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.

The data analysis in this report address the following questions:

Data Processing

This analysis makes use of dplr, knitr, reshape, xtable and ggplot2 library. Documentation of dplr can be found at http://cran.r-project.org/web/packages/dplR/dplR.pdf

# use dplr lib 
library(dplyr)
library(xtable)
library(knitr)
library(reshape)
library(ggplot2)

This analysis will use the following original variables:

and to compute dollar values for damage PROPDMGEXP and CROPDMGEXP (e.g B for billions, M for millions, etc.)

Load/Retrieve Data

# download data
setwd("~/Courses/Data Science/repos/Reproducible Research/RepData_PeerAssessment2")
dataUrl <- "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2"
dataFile <- "repdata-data-StormData.csv.bz2"

if (!file.exists(dataFile)) {
    download.file(dataUrl, dataFile, method="curl")
}
orgData <- read.csv(bzfile(dataFile))

The original data include 902297 records and 37 variables.

# select columns needed for this report
data <- orgData[,c("BGN_DATE","STATE","COUNTY","EVTYPE","FATALITIES","INJURIES","PROPDMG","PROPDMGEXP","CROPDMG","CROPDMGEXP")]

Compute dollar amount for property and crop damage

To use data for computation the values DMG columns have to be converted int dollar amounts.

# convertToDollar function will convert PROPDMGEXP or CROPDMGEXP 
# to the correct dollar amount (i.e. M for millions, B for billions, etc.)
convertToDollar <- function (x) {
    if (x == "B") {
        1e9
    } else if (x %in% c("m","M")) {
        1e6
    } else if (x %in% c("k", "K")) {
        1e3
    } else if (x %in% c("h", "H")) {
        1e2
    } else if (x %in% c("+", "-", "?")) {
        1
    } else {
        0
    }
}
# Calculate Property and Crop Damage in dollars by converting xxxxDMGEXP 
# to the dollar amount and multiplying its dollar representative
propDamage <- data$PROPDMG * unlist(lapply(data$PROPDMGEXP, function(x) convertToDollar(x)))
cropDamage <- data$CROPDMG * unlist(lapply(data$CROPDMGEXP, function(x) convertToDollar(x)))

Create Data Frame with columns needed for this analysis

# create data frame with  dollar values as number
data <- cbind(orgData[,c("BGN_DATE","STATE","COUNTY","EVTYPE","FATALITIES","INJURIES")], propDamage, cropDamage)

Compute per Event Type - Fatalities, Injuries and Damage

1. Total Fatalities, Injuries and Damage

totalFatalities <- sum(data$FATALITIES)
totalInjuries <- sum(data$INJURIES)
totalDamage <- sum(data$cropDamage + data$propDamage)
topN_perEvent <- 7
topN_State <- 10
topN_County <- 10
topN_Damage <- 10

  • total # of fatalities : 1.5145 × 104

  • total # of injuries : 1.4053 × 105

  • total damage amount : 4.7642 × 1011

2. Top 7 Fatalies per Event Type

dataByEventType <- group_by(data, EVTYPE)
eventDamage <- summarise(dataByEventType, 
    fatalities = sum(FATALITIES, na.rm = TRUE),
    injuries   = sum(INJURIES, na.rm = TRUE),
    propDamage = sum(propDamage, na.rm=TRUE),
    cropDamage = sum(cropDamage, na.rm=TRUE),
    totalDmg = sum(propDamage + cropDamage, na.rm=TRUE)
)

fatalitiesIdx <- order(eventDamage$fatalities, decreasing=TRUE)
topFatalities <- eventDamage[fatalitiesIdx[1:topN_perEvent],]

3. Top 7 Injuries per Event Type

injuryIdx <- order(eventDamage$injuries, decreasing=TRUE)
topInjury <- eventDamage[injuryIdx[1:topN_perEvent],]

Compute per State - Fatalities, Injuries and Damage

An analysis per state was to see the impact on per state level.

1. Total Fatalities, Injuries and Damage by State

by_state <- group_by(data, STATE)
state_damage <- summarise(by_state,
    fatalities = sum(FATALITIES, na.rm=TRUE),
    injuries = sum(INJURIES, na.rm=TRUE),
    propDamage  = sum(propDamage, na.rm=TRUE),
    cropDamage  = sum(cropDamage, na.rm=TRUE),
    totalDmg = sum(propDamage + cropDamage, na.rm=TRUE)    
)

2. Top 10 Fatalies per State

fatalStateIdx <- order(state_damage$fatalities, decreasing=TRUE)
topFatalState <- state_damage[fatalStateIdx[1:topN_State],]

3. Top 10 Damage in dollar per State

dmgStateIdx <- order(state_damage$totalDmg, decreasing=TRUE)
topDmgState <- state_damage[dmgStateIdx[1:topN_State],]

Compute Events with Top Damage

damageIdx <- order((eventDamage$cropDamage + eventDamage$propDamage), decreasing=TRUE)
topDollarDmg <- eventDamage[damageIdx[1:topN_Damage],]

Results

Analysis per Event Type

Top Fatalities by Event Type

print(topFatalities[,1:2], floating=FALSE)
## Source: local data frame [7 x 2]
## 
##             EVTYPE fatalities
## 834        TORNADO       5633
## 130 EXCESSIVE HEAT       1903
## 153    FLASH FLOOD        978
## 275           HEAT        937
## 464      LIGHTNING        816
## 856      TSTM WIND        504
## 170          FLOOD        470

Top Injuries by Event Type

print(topInjury[,c(1,3)])
## Source: local data frame [7 x 2]
## 
##             EVTYPE injuries
## 834        TORNADO    91346
## 856      TSTM WIND     6957
## 170          FLOOD     6789
## 130 EXCESSIVE HEAT     6525
## 464      LIGHTNING     5230
## 275           HEAT     2100
## 427      ICE STORM     1975
#kable(head(topInjury[,1:3]), format = "markdown")

Top Economic Damage by Event Type

X <- topDollarDmg[,c(1,4:6)]
X[,c(2:4)] <- X[,c(2:4)] / 1000000000
print(X)
## Source: local data frame [10 x 4]
## 
##                EVTYPE propDamage cropDamage totalDmg
## 170             FLOOD    144.658   5.661968  150.320
## 411 HURRICANE/TYPHOON     69.306   2.607873   71.914
## 834           TORNADO     56.937   0.414953   57.352
## 670       STORM SURGE     43.324   0.000005   43.324
## 244              HAIL     15.732   3.025954   18.758
## 153       FLASH FLOOD     16.141   1.421317   17.562
## 95            DROUGHT      1.046  13.972566   15.019
## 402         HURRICANE     11.868   2.741910   14.610
## 590       RIVER FLOOD      5.119   5.029459   10.148
## 427         ICE STORM      3.945   5.022113    8.967
#kable(head(X), format = "markdown")
  • Note: The numbers of damage are in billions.
X <- topDollarDmg[1:7,c(1,4:5)]
X1 <- melt(X, id=(c("EVTYPE")))
colnames(X1) <-  c("EventType","Damage","Value")
X1$Value = X1$Value / 1000000000
 
ggplot(X1, aes(x=EventType,y=Value, fill=Damage)) + 
    geom_bar(stat="identity", colour="black") +
    ggtitle("Top Damage By Event Type") +
    ylab("Damage in Billions") + xlab("Event Type") + 
    scale_fill_brewer(palette="Pastel1") +
    theme(axis.text.x = element_text(angle = 90, hjust = 1))

plot of chunk unnamed-chunk-18

Note: The numbers of propDamage and cropDamage are in billions.

Observations:

  • Tornates caused the most fatalities (over 5,000) and injuries (over 91,000).
  • Floods caused the most monitary damage, over $150 billion total wtih over $144 billion by property damage.
  • Droughts have the most negative impact on crop damage.

Analysis per State

Top Fatalities by State

topFatalState[,1:3]
## Source: local data frame [10 x 3]
## 
##    STATE fatalities injuries
## 20    IL       1421     5563
## 63    TX       1366    17667
## 51    PA        846     3223
## 2     AL        784     8742
## 37    MO        754     8998
## 13    FL        746     5918
## 38    MS        555     6675
## 8     CA        550     3278
## 5     AR        530     5550
## 62    TN        521     5202

Top Economic Damage by State

kable(head(topDmgState[,c(1,4:6)]), format = "markdown")
## 
## 
## |   |STATE | propDamage| cropDamage|  totalDmg|
## |:--|:-----|----------:|----------:|---------:|
## |8  |CA    |  1.236e+11|  3.528e+09| 1.271e+11|
## |24 |LA    |  6.007e+10|  1.229e+09| 6.130e+10|
## |13 |FL    |  4.151e+10|  3.903e+09| 4.541e+10|
## |38 |MS    |  2.981e+10|  6.610e+09| 3.642e+10|
## |63 |TX    |  2.664e+10|  7.301e+09| 3.394e+10|
## |2  |AL    |  1.724e+10|  6.068e+08| 1.785e+10|
X <- topDmgState[,c(1,4:5)]
X <- melt(X, id=(c("STATE")))
colnames(X) <-  c("State","Damage","Value")
X$Value = X$Value / 1000000000
 
ggplot(X, aes(x=State,y=Value, fill=Damage)) + 
    geom_bar(stat="identity", colour="black") +
    ggtitle("Top Damage By State") +
    ylab("Damage in Billions") + xlab("State") + 
    scale_fill_brewer(palette="Pastel1")

plot of chunk unnamed-chunk-21

kable(head(topDmgState[,c(1,6)]), format = "markdown")
## 
## 
## |   |STATE |  totalDmg|
## |:--|:-----|---------:|
## |8  |CA    | 1.271e+11|
## |24 |LA    | 6.130e+10|
## |13 |FL    | 4.541e+10|
## |38 |MS    | 3.642e+10|
## |63 |TX    | 3.394e+10|
## |2  |AL    | 1.785e+10|

Note: The numbers of propDamage and cropDamage are in billions.

Observations:

  • California has the highest damage from all states (over $127 billion), followed by Lousiana and Florida. In each case, the biggest bulk came from property damage.
  • The top states for human fatalities and injuries are Illinois, Texas, Pennsylvania, Alabama and Missouri.

Conclusion

The data analysis address the following questions:

  • 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 across the United States.