Damage Analysis on Natural Disasters based on NOAA’s Storm Database
Jiachang (Ernest) Xu
6/21/2017
Section 1: Synopsis
The objective of this project is to investigate U.S. National Oceanic and Atmospheric Administration’s (NOAA) storm database, to find which type of event has the greatest impact on population health, and poses the most severe economic consequences. We begin this analysis with loading data, and then extracting useful columns to form a clean dataset ready for further analysis. Then, we use aggregate() function to find the average fatalities, injuries, property damage, crop damage, and total damages by types of events, and extract the highest 5 types of events in each damage categories. Finally, we use the extracted top-5 data frames to plot barplots to communicate our findings.
Section 2: Data Processing
Before data cleanning, we need to load the raw data set from StormData.csv. After data loading is finished, we take a look at the top 6 rows of the raw data set
## data loading
if (!exists("storm.raw")) {
storm.raw <- read.csv("./data/StormData.csv")
}
head(storm.raw)## STATE__ BGN_DATE BGN_TIME TIME_ZONE COUNTY COUNTYNAME STATE
## 1 1 4/18/1950 0:00:00 0130 CST 97 MOBILE AL
## 2 1 4/18/1950 0:00:00 0145 CST 3 BALDWIN AL
## 3 1 2/20/1951 0:00:00 1600 CST 57 FAYETTE AL
## 4 1 6/8/1951 0:00:00 0900 CST 89 MADISON AL
## 5 1 11/15/1951 0:00:00 1500 CST 43 CULLMAN AL
## 6 1 11/15/1951 0:00:00 2000 CST 77 LAUDERDALE AL
## EVTYPE BGN_RANGE BGN_AZI BGN_LOCATI END_DATE END_TIME COUNTY_END
## 1 TORNADO 0 0
## 2 TORNADO 0 0
## 3 TORNADO 0 0
## 4 TORNADO 0 0
## 5 TORNADO 0 0
## 6 TORNADO 0 0
## COUNTYENDN END_RANGE END_AZI END_LOCATI LENGTH WIDTH F MAG FATALITIES
## 1 NA 0 14.0 100 3 0 0
## 2 NA 0 2.0 150 2 0 0
## 3 NA 0 0.1 123 2 0 0
## 4 NA 0 0.0 100 2 0 0
## 5 NA 0 0.0 150 2 0 0
## 6 NA 0 1.5 177 2 0 0
## INJURIES PROPDMG PROPDMGEXP CROPDMG CROPDMGEXP WFO STATEOFFIC ZONENAMES
## 1 15 25.0 K 0
## 2 0 2.5 K 0
## 3 2 25.0 K 0
## 4 2 2.5 K 0
## 5 2 2.5 K 0
## 6 6 2.5 K 0
## LATITUDE LONGITUDE LATITUDE_E LONGITUDE_ REMARKS REFNUM
## 1 3040 8812 3051 8806 1
## 2 3042 8755 0 0 2
## 3 3340 8742 0 0 3
## 4 3458 8626 0 0 4
## 5 3412 8642 0 0 5
## 6 3450 8748 0 0 6Since we are concerned with the relationship between types of events (EVTYPE) and population health (FATALIITIES & INJURIES) or economic consequences (PROPDMG & CROPDMG), we need to extract these 5 colmuns, and removing NA values for the purpose of data cleaning.
## extract the only 5 columns that we are interested in:
## EVTYPE, FATALITIES, INJURIES, PROPDMG, CROPDMG
storm.interested <- storm.raw[, c("EVTYPE", "FATALITIES", "INJURIES", "PROPDMG", "CROPDMG")]
## remove any rows that have NA values in any of these 5 variables
storm.clean <- storm.interested[!is.na(storm.interested$EVTYPE) &
!is.na(storm.interested$FATALITIES) &
!is.na(storm.interested$INJURIES) &
!is.na(storm.interested$PROPDMG) &
!is.na(storm.interested$CROPDMG), ]
## convert all lowercase letters to uppercase
storm.clean$EVTYPE <- toupper(storm.clean$EVTYPE)
## take a look at the top 6 rows of the clean dataset
head(storm.clean)## EVTYPE FATALITIES INJURIES PROPDMG CROPDMG
## 1 TORNADO 0 15 25.0 0
## 2 TORNADO 0 0 2.5 0
## 3 TORNADO 0 2 25.0 0
## 4 TORNADO 0 2 2.5 0
## 5 TORNADO 0 2 2.5 0
## 6 TORNADO 0 6 2.5 0Section 3: Exploratory Analysis
- Types of Events vs. Population Health
We can use aggregated() function to find the average fatalities and injuries by types of events, and store them into two new data frame. Combining these two new data frames grants us to reorder the data frame. Therefore, we can extract the top 5 types of events with highest average fatalities and injuries, separately and together. The extracted top-5 data frames are intended to be plotted in Section 4.
## find the average fatalities by types of events
storm.FATALITIES <- aggregate(FATALITIES ~ EVTYPE, storm.clean, mean)
## find the average injuries by types of events
storm.INJURIES <- aggregate(INJURIES ~ EVTYPE, storm.clean, mean)
## combine two data frame
storm.health <- cbind(storm.FATALITIES, storm.INJURIES$INJURIES)
colnames(storm.health)[3] <- "INJURIES"
## display the top 5 types of events with highest average fatalities and injuries, separately and together
storm.FATALITIES.top5 <- head(storm.FATALITIES[order(storm.FATALITIES$FATALITIES, decreasing = TRUE), ], 5)
storm.INJURIES.top5 <- head(storm.INJURIES[order(storm.INJURIES$INJURIES, decreasing = TRUE), ], 5)
storm.health.top5 <- head(storm.health[order(storm.health$FATALITIES, storm.health$INJURIES, decreasing = TRUE), ], 5)
storm.FATALITIES.top5## EVTYPE FATALITIES
## 766 TORNADOES, TSTM WIND, HAIL 25.000000
## 62 COLD AND SNOW 14.000000
## 775 TROPICAL STORM GORDON 8.000000
## 519 RECORD/EXCESSIVE HEAT 5.666667
## 127 EXTREME HEAT 4.363636storm.INJURIES.top5## EVTYPE INJURIES
## 775 TROPICAL STORM GORDON 43.0
## 872 WILD FIRES 37.5
## 746 THUNDERSTORMW 27.0
## 327 HIGH WIND AND SEAS 20.0
## 585 SNOW/HIGH WINDS 18.0storm.health.top5## EVTYPE FATALITIES INJURIES
## 766 TORNADOES, TSTM WIND, HAIL 25.000000 0.000000
## 62 COLD AND SNOW 14.000000 0.000000
## 775 TROPICAL STORM GORDON 8.000000 43.000000
## 519 RECORD/EXCESSIVE HEAT 5.666667 0.000000
## 127 EXTREME HEAT 4.363636 7.045455- Types of Events vs. Economic Consequences
We can use aggregated() function to find the average property and crop damages by types of events, and store them into two new data frame. Combining these two new data frames grants us to reorder the data frame. Therefore, we can extract the top 5 types of events with highest property, crop, and total damages on average. The extracted top-5 data frames are intended to be plotted in Section 4.
## find the average property damage by types of events
storm.PROPDMG <- aggregate(PROPDMG ~ EVTYPE, storm.clean, mean)
## find the average crop damage by types of events
storm.CROPDMG <- aggregate(CROPDMG ~ EVTYPE, storm.clean, mean)
## combine two data frame
storm.economic <- cbind(storm.PROPDMG, storm.CROPDMG$CROPDMG)
colnames(storm.economic)[3] <- "CROPDMG"
storm.economic$TOTALDMG <- storm.economic$PROPDMG + storm.economic$CROPDMG
## extract the top 5 type of events with highest average property and crop damages, seperately and together
storm.PROPDMG.top5 <- head(storm.PROPDMG[order(storm.PROPDMG$PROPDMG, decreasing = TRUE), ], 5)
storm.CROPDMG.top5 <- head(storm.CROPDMG[order(storm.CROPDMG$CROPDMG, decreasing = TRUE), ], 5)
storm.economic.top5 <- head(storm.economic[order(storm.economic$TOTALDMG, decreasing = TRUE), ], 5)
## display the top 5 type of events with highest average property and crop damages, seperately and together
storm.PROPDMG.top5## EVTYPE PROPDMG
## 48 COASTAL EROSION 766
## 255 HEAVY RAIN AND FLOOD 600
## 528 RIVER AND STREAM FLOOD 600
## 36 BLIZZARD/WINTER STORM 500
## 143 FLASH FLOOD/ 500storm.CROPDMG.top5## EVTYPE CROPDMG
## 106 DUST STORM/HIGH WINDS 500
## 173 FOREST FIRES 500
## 775 TROPICAL STORM GORDON 500
## 353 HIGH WINDS/COLD 401
## 367 HURRICANE FELIX 250storm.economic.top5## EVTYPE PROPDMG CROPDMG TOTALDMG
## 775 TROPICAL STORM GORDON 500 500 1000
## 48 COASTAL EROSION 766 0 766
## 255 HEAVY RAIN AND FLOOD 600 0 600
## 528 RIVER AND STREAM FLOOD 600 0 600
## 106 DUST STORM/HIGH WINDS 50 500 550Section 4: Results
- Types of Events vs. Population Health
The data frames that store the top 5 types of events with highest average fatalities and injuries are already extracted in Section 3. We can plot them, using barplot in descending order.
require(ggplot2)## Loading required package: ggplot2require(gridExtra)## Loading required package: gridExtra## reorder
storm.FATALITIES.top5 <- transform(storm.FATALITIES.top5, EVTYPE = reorder(EVTYPE, -FATALITIES))
storm.INJURIES.top5 <- transform(storm.INJURIES.top5, EVTYPE = reorder(EVTYPE, -INJURIES))
plot1.1 <- ggplot(data = storm.FATALITIES.top5, aes(x = EVTYPE, y = FATALITIES)) +
geom_bar(stat = "identity", fill = "firebrick") +
geom_text(aes(label = as.integer(FATALITIES)), vjust = 1.6, color = "black", size = 3.5) +
xlab("Types of Events") + ylab("Fatalities") +
ggtitle("Figure 1.1: Top 5 Types of Events with Highest Fatalities")
plot1.2 <- ggplot(data = storm.INJURIES.top5, aes(x = EVTYPE, y = INJURIES)) +
geom_bar(stat = "identity", fill = "orange3") +
geom_text(aes(label = as.integer(INJURIES)), vjust = 1.6, color = "black", size = 3.5) +
xlab("Types of Events") + ylab("Injuries") +
ggtitle("Figure 1.2: Top 5 Types of Events with Highest Injuries")
grid.arrange(plot1.1, plot1.2, ncol=1)
Figure 1.1 shows that TORNADOES, TSTM WIND, HAIL has the highest average fatalities across all types of events; Figure 1.2 shows that TROPICAL STORM GORDON has the highest average injuries across all types of events.
- Types of Events vs. Economic Consequences
The data frames that store the top 5 types of events with highest property, crop, and total damages on average are already extracted in Section 3. We can plot them, using barplot in descending order.
require(ggplot2)
require(gridExtra)
## reorder
storm.PROPDMG.top5 <- transform(storm.PROPDMG.top5, EVTYPE = reorder(EVTYPE, -PROPDMG))
storm.CROPDMG.top5 <- transform(storm.CROPDMG.top5, EVTYPE = reorder(EVTYPE, -CROPDMG))
storm.economic.top5 <- transform(storm.economic.top5, EVTYPE = reorder(EVTYPE, -TOTALDMG))
plot2.1 <- ggplot(data = storm.PROPDMG.top5, aes(x = EVTYPE, y = PROPDMG)) +
geom_bar(stat = "identity", fill = "olivedrab") +
geom_text(aes(label = as.integer(PROPDMG)), vjust = 1.6, color = "black", size = 3.5) +
xlab("Types of Events") + ylab("Property Damage") +
ggtitle("Figure 2.1: Top 5 Types of Events with Highest Property Damage")
plot2.2 <- ggplot(data = storm.CROPDMG.top5, aes(x = EVTYPE, y = CROPDMG)) +
geom_bar(stat = "identity", fill = "steelblue") +
geom_text(aes(label = as.integer(CROPDMG)), vjust = 1.6, color = "black", size = 3.5) +
xlab("Types of Events") + ylab("Crop Damage") +
ggtitle("Figure 2.2: Top 5 Types of Events with Highest Crop Damages")
plot2.3 <- ggplot(data = storm.economic.top5, aes(x = EVTYPE, y = TOTALDMG)) +
geom_bar(stat = "identity", fill = "blueviolet") +
geom_text(aes(label = as.integer(TOTALDMG)), vjust = 1.6, color = "black", size = 3.5) +
xlab("Types of Events") + ylab("Total Damages") +
ggtitle("Figure 2.3: Top 5 Types of Events with Highest Total Damages")
grid.arrange(plot2.1, plot2.2, plot2.3, nrow=3)
Figure 2.1 shows that COASTAL EROSION has the highest average property damage across all types of events; Figure 2.2 shows that DUST STORM/HIGH WINDS has the highest average crop damage across all type of events; Figure 2.3 shows that TROPICAL STORM GORDON has the highest total damages on average.