Questions:

Which types of events (as indicated in the EVTYPE variable) across the United States:

are most harmful with respect to population health, across the United States?
have the greatest economic consequences?

Requirements:

Publish to RPubs.
Compile to HTML.
Doc layout:

Language: English.
Title: summarizes your data analysis
Synopsis section: describes and summarizes the analysis in at most 10 complete sentences.
Data Processing section: code and narration
Result section
Figures: max of three figures.

1. Synopsis

This is my peer assignment 2 for Coursera: Reproducible Research. This document is my own work, based on several examples of workflow that I found in the Rpubs and StackOverflow site. One of the main reference is Ryan Cheung Git.

The following are the results:

Injuries:
- Tornado is the cause for most injuries (66%).
- tornado, wind and heat are the top three weather events for over 80% injuries.
Fatalities:
- Tornado is the cause for most fatalities (38.4%).
- Tornado, heat, and flood are the top three weather events for about 70% fatalities.
Porperty Damage: Flood is the cause for most porperty damages(48%) among all weather events. Flood and tornado together are responsible for about 65% of all porperty damages.
Crop Damage: Drought is the cause for most crop damages(31%) among all weather events. Drought, flood and ice storm together are responsible for over 69% of all crop damages.

2. Data processing

2.1 Preparation (set wd, load libraries, read data)

library(ggplot2)
library(scales)

data <- read.csv("repdata-data-StormData.csv", header = T)
names(data) <- tolower(names(data))
names(data) <- gsub("_","",names(data))

2.2 Clean the values in `EVTYPE` column

According to the dataset documentation the event name should be one of those fifty listed. Here I use tolower() function.

data$evtype <- tolower(as.character(data$evtype))
data$evtype <- gsub("^(([^:]+)://)?([^:/]+)(:([0-9]+))?(/.*)","",data$evtype)

2.3 Removing unwanted values

Too small values

data <- subset(data,nchar(data$evtype)>=2)

Values starts with “summary”

data$evtype[grep("summary", data$evtype)] <- "tbm"
data <- subset(data,data$evtype != "tbm")

2.4 Consolidating values

Using grep() function to the dataset

data$evtype[grep("hail", data$evtype)] <- "Hail"
data$evtype[grep("wind", data$evtype)] <- "Wind"
data$evtype[grep("tornado", data$evtype)] <- "Tornado"
data$evtype[grep("flood", data$evtype)] <- "Flood"
data$evtype[grep("lightning", data$evtype)] <- "Lightning"
data$evtype[grep("snow", data$evtype)] <- "Snow"
data$evtype[grep("rain", data$evtype)] <- "Rain"
data$evtype[grep("winter", data$evtype)] <- "Winter"
data$evtype[grep("heat", data$evtype)] <- "Heat"
data$evtype[grep("fog", data$evtype)] <- "Fog"
data$evtype[grep("surf", data$evtype)] <- "Surf"
data$evtype[grep("ice storm", data$evtype)] <- "Ice storm"
data$evtype[grep("fire", data$evtype)] <- "Wild fire"
data$evtype[grep("storm surge", data$evtype)] <- "Strom surge"
data$evtype[grep("hurricane", data$evtype)] <- "Hurricane"
data$evtype[grep("drought", data$evtype)] <- "Drought"
data$evtype[grep("thunderstorm", data$evtype)] <- "Thunderstorm"

The following major weather event types takes up 97% of all observations.
The rest 3% weather events are changed into new category named “other”.

sum(data$evtype %in% c("Flood","Wind","Snow",
                       "Tornado","Hail","Rain",
                       "Lightning","Winter","Fog",
                       "Heat","Surf","Ice storm",
                       "Wild fire","Storm surge",
                       "Hurricane","Drought",
                       "Thunderstorm"))/nrow(data)

## [1] 0.9778

tbc <- data$evtype %in% c("Flood","Wind","Snow",
                          "Tornado","Hail","Rain",
                          "Lightning","Winter","Heat",
                          "Surf","Fog","Ice storm",
                          "Wild fire","Storm surge",
                          "Hurricane","Drought",
                          "Thunderstorm") == F
                          data$evtype[tbc == T] <- "other"

Summary of counts of cleaned event type values.

sort(table(data$evtype))

## 
## Thunderstorm    Hurricane         Surf          Fog    Ice storm 
##           92          199          833         1883         2006 
##      Drought         Heat    Wild fire         Rain    Lightning 
##         2488         2630         2781        12136        15762 
##         Snow       Winter        other      Tornado        Flood 
##        17569        18492        19813        60688        81967 
##         Hail         Wind 
##       289338       362164

2.5 Calculating Damages

A. Preparation

data$propdmgexp <- as.character(data$propdmgexp)
data$propdmgexp[grep("K", data$propdmgexp)] <- "1000"
data$propdmgexp[grep("M", data$propdmgexp)] <- "1000000"
data$propdmgexp[grep("m", data$propdmgexp)] <- "1000000"
data$propdmgexp[grep("B", data$propdmgexp)] <- "1000000000"
tbc <- data$propdmgexp %in% c("1000","1000000","1000000000") == F
data$propdmgexp[tbc == T] <- "1"
data$propdmgexp <- as.numeric(data$propdmgexp)

data$cropdmgexp <- as.character(data$cropdmgexp)
data$cropdmgexp[grep("K", data$cropdmgexp)] <- "1000"
data$cropdmgexp[grep("M", data$cropdmgexp)] <- "1000000"
data$cropdmgexp[grep("m", data$cropdmgexp)] <- "1000000"
data$cropdmgexp[grep("B", data$cropdmgexp)] <- "1000000000"
tbc <- data$cropdmgexp %in% c("1000","1000000","1000000000") == F
data$cropdmgexp[tbc == T] <- "1"
data$cropdmgexp <- as.numeric(data$cropdmgexp)

B. Calculating and stored damages as new variable “propDamage” and “cropDamage”

data$propdamage <- data$propdmg * data$propdmgexp
data$cropdamage <- data$cropdmg * data$cropdmgexp

3. Results

3.1 Damages to Population Health

Injuries:
- Tornado is the cause for most injuries (66%).
- tornado, wind and heat are the top three weather events for over 80% injuries.
Fatalities:
- Tornado is the cause for most fatalities (38%).
- Tornado, heat, and flood are the top three weather events for about 70% fatalities.

A. The event that caused most injuries.

totalInjuries <- tapply(data$injuries, data$evtype, sum)
sort(totalInjuries, decreasing = T)[1]

## Tornado 
##   91365

B. The percentage of injuries caused by the tornado.

sum(sort(totalInjuries, decreasing = T)[1])/sum(totalInjuries)

## [1] 0.661

C. The top 3 events responsible for most injuries

sort(totalInjuries, decreasing = T)[1:3]

## Tornado    Wind    Heat 
##   91365   11319    9224

D. The percentage of injuries caused by the top 3 weather events

sum(sort(totalInjuries, decreasing = T)[1:3])/sum(totalInjuries)

## [1] 0.8096

E. The event that caused most fatalities

totalFatal <- tapply(data$fatalities, data$evtype, sum)
sort(totalFatal, decreasing = T)[1]

## Tornado 
##    5633

F. The percentage of fatalities caused by the tornado

sum(sort(totalFatal, decreasing = T)[1])/sum(totalFatal)

## [1] 0.3846

G. The top 3 events responsible for most fatalities

sort(totalFatal, decreasing = T)[1:3]

## Tornado    Heat   Flood 
##    5633    3119    1488

H. The percentage of death caused by the top 3 weather events

sum(sort(totalFatal, decreasing = T)[1:3])/sum(totalFatal)

## [1] 0.6992

3.2 Economic Consequences

Porperty Damage: Flood is the cause for most porperty damages(48%) among all weather events. Flood and tornado together are responsible for about 65% of all porperty damages.
Crop Damage: Drought is the cause for most crop damages(31%) among all weather events. Drought, flood and ice storm together are responsible for over 69% of all crop damages.

A. The event that caused most prop damages

totalPropDamage <- tapply(data$propdamage, data$evtype, sum)
sort(totalPropDamage, decreasing = T)[1]

##    Flood 
## 1.67e+11

B. The percentage of prop damages caused by flood

sort(totalPropDamage, decreasing = T)[1]/sum(totalPropDamage)

##  Flood 
## 0.4814

C.Top 2 weather events causing the most prop damages

sort(totalPropDamage, decreasing = T)[1:2]

##     Flood   Tornado 
## 1.670e+11 5.694e+10

D. Percentage of total prop damages caused by the top 2 weather events

sum(sort(totalPropDamage, decreasing = T)[1:2])/sum(totalPropDamage)

## [1] 0.6456

E. The event that caused most crop damages

totalCropDamage <- tapply(data$cropdamage, data$evtype, sum)
sort(totalCropDamage, decreasing = T)[1]

##   Drought 
## 1.397e+10

F. The percentage of crop damages caused by drought

sort(totalCropDamage, decreasing = T)[1]/sum(totalCropDamage)

## Drought 
##  0.3106

G. Top 3 weather events causing the most crop damages

sort(totalCropDamage, decreasing = T)[1:3]

##   Drought     Flood Ice storm 
## 1.397e+10 1.217e+10 5.022e+09

H. Percentage of total porperty damages caused by the top 3 weather events

sum(sort(totalCropDamage, decreasing = T)[1:3])/sum(totalCropDamage)

## [1] 0.6927

Figure 1 Barplot of damages to population health

plot.new()
par(mfrow=c(1,2))
barplot(sort(totalInjuries, 
             decreasing = T), 
             main = "Number of injuries")

barplot(sort(totalFatal, 
             decreasing = T), 
             main = "Number of fatalities")

plot of chunk unnamed-chunk-26

Figure 2 Barplot of weather events to economic consequences

par(mfrow=c(1,2))
barplot(sort(totalPropDamage, 
             decreasing = T), 
             main = "Number of property damage")

barplot(sort(totalCropDamage, 
             decreasing = T), main = "Number of crop damages")

plot of chunk unnamed-chunk-27

References

Weather data exploration: Coursera-PeerAssignment02

Dasapta Erwin Irawan

15/09/2014

Questions:

Requirements:

1. Synopsis

2. Data processing

2.1 Preparation (set wd, load libraries, read data)

2.2 Clean the values in `EVTYPE` column

2.3 Removing unwanted values

2.4 Consolidating values

2.5 Calculating Damages

A. Preparation

B. Calculating and stored damages as new variable “propDamage” and “cropDamage”

3. Results

3.1 Damages to Population Health

A. The event that caused most injuries.

B. The percentage of injuries caused by the tornado.

C. The top 3 events responsible for most injuries

D. The percentage of injuries caused by the top 3 weather events

E. The event that caused most fatalities

F. The percentage of fatalities caused by the tornado

G. The top 3 events responsible for most fatalities

H. The percentage of death caused by the top 3 weather events

3.2 Economic Consequences

A. The event that caused most prop damages

B. The percentage of prop damages caused by flood

C.Top 2 weather events causing the most prop damages

D. Percentage of total prop damages caused by the top 2 weather events

E. The event that caused most crop damages

F. The percentage of crop damages caused by drought

G. Top 3 weather events causing the most crop damages

H. Percentage of total porperty damages caused by the top 3 weather events

Weather data exploration: Coursera-PeerAssignment02

Dasapta Erwin Irawan

15/09/2014

Questions:

Requirements:

1. Synopsis

2. Data processing

2.1 Preparation (set wd, load libraries, read data)

2.2 Clean the values in EVTYPE column

2.3 Removing unwanted values

2.4 Consolidating values

2.5 Calculating Damages

A. Preparation

B. Calculating and stored damages as new variable “propDamage” and “cropDamage”

3. Results

3.1 Damages to Population Health

A. The event that caused most injuries.

B. The percentage of injuries caused by the tornado.

C. The top 3 events responsible for most injuries

D. The percentage of injuries caused by the top 3 weather events

E. The event that caused most fatalities

F. The percentage of fatalities caused by the tornado

G. The top 3 events responsible for most fatalities

H. The percentage of death caused by the top 3 weather events

3.2 Economic Consequences

A. The event that caused most prop damages

B. The percentage of prop damages caused by flood

C.Top 2 weather events causing the most prop damages

D. Percentage of total prop damages caused by the top 2 weather events

E. The event that caused most crop damages

F. The percentage of crop damages caused by drought

G. Top 3 weather events causing the most crop damages

H. Percentage of total porperty damages caused by the top 3 weather events

2.2 Clean the values in `EVTYPE` column