Cathastrophic event analysis by type

Synopsis

This document contains 3 elements:

• description of transformations that were made to the raw dataset in order to obtain final datasets used in the analysis
• answer to the question of which catastrophic event type is the most impactful when it comes to human health
• answer to the question of which catastrophic event type is the most impactful when it comes to economic situation

Data procesing

In order to determine the most impactful type of cathastropic event when it comes to both human health as well as economic situation I had to design new measures. I could not use a simple sum of measure like fatalities because some events occured more frequently and therefore those would have higher value of sum. In sted of that I calculated measure per event so i divided the sum of for ex. fatalities for a particular event by the nr of times this event happend. This way I got a measure that allowed for a comparison of different events.

setwd("C:/Users/T540pDLEWYNBQ/Google Drive/Inne/Coursera/Reproducible Research/Assignment 2")
stormdata<-read.csv("FStormData.bz2")

library(dplyr)
FATALITIES_BY_EVTYPE<-stormdata %>% 
    group_by(EVTYPE) %>% 
    summarise("count"=n(),
              "FATALITIES_SUM"=sum(FATALITIES, na.rm=T)) %>% 
    mutate("FATALITIES_PER_EVENT"=FATALITIES_SUM/count) %>% 
    arrange(FATALITIES_PER_EVENT) %>% 
    top_n(10,FATALITIES_PER_EVENT)
INJURIES_BY_EVTYPE<-stormdata %>% 
    group_by(EVTYPE) %>% 
    summarise("count"=n(),
              "INJURIES_SUM"=sum(INJURIES, na.rm=T)) %>% 
    mutate("INJURIES_PER_EVENT"=INJURIES_SUM/count) %>% 
    arrange(INJURIES_PER_EVENT) %>% 
    top_n(10,INJURIES_PER_EVENT)
names<-FATALITIES_BY_EVTYPE$EVTYPE
FATALITIES_BY_EVTYPE$EVTYPE<-factor(FATALITIES_BY_EVTYPE$EVTYPE, levels = names)
names<-INJURIES_BY_EVTYPE$EVTYPE
INJURIES_BY_EVTYPE$EVTYPE<-factor(INJURIES_BY_EVTYPE$EVTYPE, levels = names)
PROPDMG_BY_EVTYPE<-stormdata %>% 
    group_by(EVTYPE) %>% 
    summarise("count"=n(),
              "PROPDMG_SUM"=sum(PROPDMG, na.rm=T)) %>% 
    mutate("PROPDMG_PER_EVENT"=PROPDMG_SUM/count) %>% 
    arrange(PROPDMG_PER_EVENT) %>% 
    top_n(10,PROPDMG_PER_EVENT)
CROPDMG_BY_EVTYPE<-stormdata %>% 
    group_by(EVTYPE) %>% 
    summarise("count"=n(),
              "CROPDMG_SUM"=sum(CROPDMG, na.rm=T)) %>% 
    mutate("CROPDMG_PER_EVENT"=CROPDMG_SUM/count) %>% 
    arrange(CROPDMG_PER_EVENT) %>% 
    top_n(10,CROPDMG_PER_EVENT)
names<-PROPDMG_BY_EVTYPE$EVTYPE
PROPDMG_BY_EVTYPE$EVTYPE<-factor(PROPDMG_BY_EVTYPE$EVTYPE, levels = names)
names<-CROPDMG_BY_EVTYPE$EVTYPE
CROPDMG_BY_EVTYPE$EVTYPE<-factor(CROPDMG_BY_EVTYPE$EVTYPE, levels = names)

Results - Question 1

Below you can find the list of top 10 most impactful catastrophic events when it comes to human fatalities and injuries.

library(ggplot2)
a<-ggplot(data=FATALITIES_BY_EVTYPE)+
    geom_bar(aes(x=EVTYPE, y = FATALITIES_PER_EVENT), stat = "identity")+
    coord_flip()+
    ggtitle("FATALITIES PER EVENT")
b<-ggplot(data=INJURIES_BY_EVTYPE)+
    geom_bar(aes(x=EVTYPE, y = INJURIES_PER_EVENT), stat = "identity")+
    coord_flip()+
    ggtitle("INJURIES PER EVENT")
multiplot(a,b, cols=2)

Results - Question 2

Below you can find the list of top 10 most impactful catastrophic events when it comes to propery damage and crop damage.

library(ggplot2)
a<-ggplot(data=PROPDMG_BY_EVTYPE)+
    geom_bar(aes(x=EVTYPE, y = PROPDMG_PER_EVENT), stat = "identity")+
    coord_flip()+
    ggtitle("PROPERTY DAMAGE PER EVENT")
b<-ggplot(data=CROPDMG_BY_EVTYPE)+
    geom_bar(aes(x=EVTYPE, y = CROPDMG_PER_EVENT), stat = "identity")+
    coord_flip()+
    ggtitle("CROP DAMAGE PER EVENT")
multiplot(a,b, cols=2)