Synopsis

This report analyzes the U.S. National Oceanic and Atmospheric Administration (NOAA) Storm Database (raw CSV.bz2) to answer two questions: (1) which types of events are most harmful to population health (fatalities + injuries), and (2) which event types have the greatest economic consequences (property + crop damage). The analysis begins from the raw compressed CSV file, performs necessary data cleaning and unit conversion for damage exponents, and aggregates impacts by EVTYPE. Results are presented with reproducible R code (using dplyr and ggplot2) and two figures: a top-10 event types by health impact and a top-10 by economic impact. All code chunks show their source (echo = TRUE) so the work is reproducible. The document includes a Data Processing section describing transformations and a Results section with figures and interpretation.

Data Processing

This section loads the raw data, documents assumptions and transformations, and prepares the summary tables used in the Results.

knitr::opts_chunk$set(echo = TRUE, message = TRUE, warning = TRUE)
library(dplyr)

## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

library(ggplot2)
library(readr)

Download and read raw data

The analysis starts from the original compressed CSV file provided for the assignment. If you already have the file locally, set local_file accordingly. The default below downloads from the course URL (the standard link used in the Coursera assignment site).

# URL used in the original assignment (standard location for the Coursera assignment)
url <- "https://d396qusza40orc.cloudfront.net/repdata%2Fdata%2FStormData.csv.bz2"
local_file <- "StormData.csv.bz2"

if (!file.exists(local_file)) {
  download.file(url, destfile = local_file, mode = "wb")
}

# read directly from the bzip2 compressed file
storms <- read.csv(local_file, stringsAsFactors = FALSE)

# quick peek
dim(storms)

## [1] 902297     37

str(storms[c("EVTYPE", "FATALITIES", "INJURIES", "PROPDMG", "PROPDMGEXP", "CROPDMG", "CROPDMGEXP")])

## 'data.frame':    902297 obs. of  7 variables:
##  $ EVTYPE    : chr  "TORNADO" "TORNADO" "TORNADO" "TORNADO" ...
##  $ FATALITIES: num  0 0 0 0 0 0 0 0 1 0 ...
##  $ INJURIES  : num  15 0 2 2 2 6 1 0 14 0 ...
##  $ PROPDMG   : num  25 2.5 25 2.5 2.5 2.5 2.5 2.5 25 25 ...
##  $ PROPDMGEXP: chr  "K" "K" "K" "K" ...
##  $ CROPDMG   : num  0 0 0 0 0 0 0 0 0 0 ...
##  $ CROPDMGEXP: chr  "" "" "" "" ...

Notes on variables and transformations

EVTYPE is the event type (character). Many entries use inconsistent capitalization or spelling; for a first-pass analysis we will use simple standardization (trim, upper-case) and then aggregate by the cleaned EVTYPE values. A more thorough analysis could manually collapse similar event names into canonical categories.
FATALITIES and INJURIES are numeric counts and can be summed directly.
PROPDMG and CROPDMG represent numeric amounts with multipliers indicated in PROPDMGEXP / CROPDMGEXP. The exponent field contains characters like K, M, B, numeric digits, or other symbols. We convert these to multipliers and compute total dollar damages.

# helper: convert exponent code to multiplier
exp_to_mult <- function(e) {
  e <- toupper(trimws(e))
  e[e == ""] <- ""
  sapply(e, function(x) {
    if (x %in% c("H")) return(100)
    if (x %in% c("K")) return(1e3)
    if (x %in% c("M")) return(1e6)
    if (x %in% c("B")) return(1e9)
    # sometimes numeric characters "0"-"9" indicate power of ten
    if (grepl("^[0-9]+$", x)) return(10^as.numeric(x))
    # + or - or ? or other characters: treat as multiplier 1 (conservative)
    return(1)
  })
}

storms <- storms %>%
  mutate(
    EVTYPE_CLEAN = toupper(trimws(EVTYPE)),
    PROPDMGEXP_CLEAN = toupper(trimws(PROPDMGEXP)),
    CROPDMGEXP_CLEAN = toupper(trimws(CROPDMGEXP))
  )

# compute multiplier and dollar damage
storms <- storms %>%
  mutate(
    prop_mult = exp_to_mult(PROPDMGEXP_CLEAN),
    crop_mult = exp_to_mult(CROPDMGEXP_CLEAN),
    PROPDMG_USD = as.numeric(PROPDMG) * prop_mult,
    CROPDMG_USD = as.numeric(CROPDMG) * crop_mult,
    HEALTH_IMPACT = as.numeric(FATALITIES) + as.numeric(INJURIES),
    ECONOMIC_IMPACT = PROPDMG_USD + CROPDMG_USD
  )

# quick sanity checks
summary(storms$PROPDMG_USD)

##      Min.   1st Qu.    Median      Mean   3rd Qu.      Max. 
## 0.000e+00 0.000e+00 0.000e+00 4.746e+05 5.000e+02 1.150e+11

summary(storms$CROPDMG_USD)

##      Min.   1st Qu.    Median      Mean   3rd Qu.      Max. 
## 0.000e+00 0.000e+00 0.000e+00 5.442e+04 0.000e+00 5.000e+09

Aggregation strategy

For population health: sum FATALITIES + INJURIES per EVTYPE_CLEAN.
For economic impact: sum PROPDMG_USD + CROPDMG_USD per EVTYPE_CLEAN.
We’ll present the top 10 event types for each metric.

health_by_event <- storms %>%
  group_by(EVTYPE_CLEAN) %>%
  summarise(
    total_fatalities = sum(FATALITIES, na.rm = TRUE),
    total_injuries = sum(INJURIES, na.rm = TRUE),
    total_health = sum(HEALTH_IMPACT, na.rm = TRUE),
    n_events = n()
  ) %>%
  arrange(desc(total_health))

## `summarise()` ungrouping output (override with `.groups` argument)

econ_by_event <- storms %>%
  group_by(EVTYPE_CLEAN) %>%
  summarise(
    total_prop = sum(PROPDMG_USD, na.rm = TRUE),
    total_crop = sum(CROPDMG_USD, na.rm = TRUE),
    total_econ = sum(ECONOMIC_IMPACT, na.rm = TRUE),
    n_events = n()
  ) %>%
  arrange(desc(total_econ))

## `summarise()` ungrouping output (override with `.groups` argument)

# top 10
top10_health <- head(health_by_event, 10)
top10_econ <- head(econ_by_event, 10)

knitr::kable(top10_health, caption = "Top 10 event types by combined injuries+fatalities")

Top 10 event types by combined injuries+fatalities
EVTYPE_CLEAN	total_fatalities	total_injuries	total_health	n_events
TORNADO	5633	91346	96979	60652
EXCESSIVE HEAT	1903	6525	8428	1678
TSTM WIND	504	6957	7461	219946
FLOOD	470	6789	7259	25327
LIGHTNING	816	5230	6046	15755
HEAT	937	2100	3037	767
FLASH FLOOD	978	1777	2755	54278
ICE STORM	89	1975	2064	2006
THUNDERSTORM WIND	133	1488	1621	82564
WINTER STORM	206	1321	1527	11433

knitr::kable(top10_econ, caption = "Top 10 event types by combined property+crop damage (USD)")

Top 10 event types by combined property+crop damage (USD)
EVTYPE_CLEAN	total_prop	total_crop	total_econ	n_events
FLOOD	144657709807	5661968450	150319678257	25327
HURRICANE/TYPHOON	69305840000	2607872800	71913712800	88
TORNADO	56947380676	414953270	57362333946	60652
STORM SURGE	43323536000	5000	43323541000	261
HAIL	15735267513	3025954473	18761221986	288661
FLASH FLOOD	16822723978	1421317100	18244041078	54278
DROUGHT	1046106000	13972566000	15018672000	2488
HURRICANE	11868319010	2741910000	14610229010	174
RIVER FLOOD	5118945500	5029459000	10148404500	173
ICE STORM	3944927860	5022113500	8967041360	2006

Results

This section presents the plots and interprets findings. Figures are limited to two (health and economic), which keeps within the assignment limit of at most three figures.

Figure 1 — Which events are most harmful to population health?

# prepare data for plotting
plot_health <- top10_health %>%
  mutate(EVTYPE_CLEAN = reorder(EVTYPE_CLEAN, total_health))

ggplot(plot_health, aes(x = EVTYPE_CLEAN, y = total_health)) +
  geom_col() +
  coord_flip() +
  labs(x = "Event type", y = "Total fatalities + injuries",
       title = "Top 10 NOAA Event Types by Health Impact") +
  theme_minimal()

Top 10 event types by total fatalities+injuries (descending).

Caption: This bar chart shows the top 10 event types ranked by combined fatalities and injuries recorded in the NOAA Storm Database (1950–2011). Interpretation notes: event naming in the raw data is inconsistent; some categories (e.g., variations of thunderstorms, hail, wind) may be split across multiple EVTYPE strings and require manual cleaning for final policy use.

Figure 2 — Which events have the greatest economic consequences?

plot_econ <- top10_econ %>%
  mutate(EVTYPE_CLEAN = reorder(EVTYPE_CLEAN, total_econ))

# plot with log scale to manage wide range of damages
ggplot(plot_econ, aes(x = EVTYPE_CLEAN, y = total_econ)) +
  geom_col() +
  coord_flip() +
  scale_y_continuous(labels = scales::dollar_format(prefix = "$", scale = 1, big.mark = ",")) +
  labs(x = "Event type", y = "Total damage (USD)",
       title = "Top 10 NOAA Event Types by Economic Impact (Property + Crop)") +
  theme_minimal()

Top 10 event types by economic damage (property + crop), in US dollars.

Caption: The bar chart uses absolute USD to show total property and crop damage by event type. Note that damage amounts span several orders of magnitude; for deeper analysis consider plotting on a log scale or presenting median per-event damage.

Interpretation and Limitations

The results identify which event types have historically caused the most harm and cost. These are presented as ranked lists and bar charts. Use these findings as a starting point for resource prioritization.
Limitations: EVTYPE entries are noisy — the same phenomenon may appear under many spellings (THUNDERSTORM WIND, TSTM WIND, THUNDERSTORM WINDS, etc.). A thorough policy-grade analysis must clean and map EVTYPE to canonical categories (the NOAA documentation suggests an event type standardization list). The exponent fields for damage contain many nonstandard characters and ambiguous entries; our conservative choice treats unknown symbols as multiplier 1. This may under- or over-estimate some totals.
Extensions: collapse EVTYPE to canonical categories; compute per-event averages; separate analysis by decade; map results geographically.

Reproducibility

All code shown above starts from the raw StormData.csv.bz2 file and uses open R packages only. To reproduce the figures and tables, open this R Markdown document in RStudio and knit it to HTML (it will download the raw file automatically if not present).

Appendix — Full code listing

(The full code is shown inline in the document; all chunks have echo = TRUE.)

End of report.

NOAA Storm Database — Which Events Harm Health and Economy Most?

Your Name

2025-10-18