World wide natural disaster events — Analysis Report

Author

Ziaul Hoque

Published

September 29, 2026

Approach

In this assignment, I selected a dataset from kaggle, regarding an analytical investigation of worldwide natural disaster occurrences documented between 2023 and 2025. This dataset contains important characteristics such the type of disaster, location, degree of severity, number of victims, financial losses, and emergency response timings are all captured here and also gives an opportunity to working with enormous numerical data.

My goal is to work with this data and get insights information through data driven approach and analyze incidents in different locations worldwide and get more information & visualize data to improve emergency response plans, risk mitigation techniques and disaster preparedness. So data driven insights are essential for identifying susceptible areas and allocating resources as emergency needed.

Data Sources

https://github.com/CandySoftInc/DATA607-Assignment1-Natural-Disaster-Events/blob/main/synthetic_disaster_events_2025.csv

Research Question

How we can improve more awareness to people before any natural disaster happens and how we can prevent loss or degree of severity, number of victims through this analysis?

Codebase

Including all related library

Code

library(tidyverse)
library(lubridate)
library(scales)
library(knitr)

Import Dataset

Loading dataset into dataframe to clean and preparing a dataset by modifying its columns into correct datatypes.

Code

data_file <- "synthetic_disaster_events_2025.csv"
data <- read_csv(data_file, show_col_types = FALSE) |>
  mutate(
    date = as.Date(date),
    disaster_type = factor(disaster_type),
    location = factor(location),
    aid_provided = factor(aid_provided, levels = c("No", "Yes")),
    is_major_disaster = as.logical(is_major_disaster)
  )

cat("Rows:", nrow(data), "  Columns:", ncol(data))

Rows: 20000   Columns: 13

Check Dataset Structure

Code

str(data)

tibble [20,000 × 13] (S3: tbl_df/tbl/data.frame)
 $ event_id                   : num [1:20000] 1 2 3 4 5 6 7 8 9 10 ...
 $ disaster_type              : Factor w/ 7 levels "Drought","Earthquake",..: 7 4 6 1 6 7 1 6 4 1 ...
 $ location                   : Factor w/ 8 levels "Chile","India",..: 1 2 4 1 7 3 2 5 7 8 ...
 $ latitude                   : num [1:20000] -34.7 22.1 42.3 -33.4 39.4 ...
 $ longitude                  : num [1:20000] -71.8 78 11 -70 37 ...
 $ date                       : Date[1:20000], format: "2025-08-27" "2023-05-29" ...
 $ severity_level             : num [1:20000] 8 5 7 8 8 9 7 7 4 4 ...
 $ affected_population        : num [1:20000] 31104 29340 34804 31191 46284 ...
 $ estimated_economic_loss_usd: num [1:20000] 2768213 5996227 9222541 1827703 13435921 ...
 $ response_time_hours        : num [1:20000] 5.12 44.43 49.3 65.56 60.96 ...
 $ aid_provided               : Factor w/ 2 levels "No","Yes": 2 1 1 2 1 2 2 2 1 2 ...
 $ infrastructure_damage_index: num [1:20000] 0.59 0.26 0.94 0.94 0.92 0.95 0.4 1 0.53 0.64 ...
 $ is_major_disaster          : logi [1:20000] TRUE FALSE TRUE TRUE TRUE TRUE ...

Code

head(data)

# A tibble: 6 × 13
  event_id disaster_type   location latitude longitude date       severity_level
     <dbl> <fct>           <fct>       <dbl>     <dbl> <date>              <dbl>
1        1 Wildfire        Chile      -34.7      -71.8 2025-08-27              8
2        2 Hurricane       India       22.1       78.0 2023-05-29              5
3        3 Volcanic Erupt… Italy       42.3       11.0 2023-01-15              7
4        4 Drought         Chile      -33.4      -70.0 2024-02-08              8
5        5 Volcanic Erupt… Turkey      39.4       37.0 2023-12-23              8
6        6 Wildfire        Indones…    -2.48     113.  2023-11-19              9
# ℹ 6 more variables: affected_population <dbl>,
#   estimated_economic_loss_usd <dbl>, response_time_hours <dbl>,
#   aid_provided <fct>, infrastructure_damage_index <dbl>,
#   is_major_disaster <lgl>

Dataset Overview

Code

overview <- tibble(
  Measure = c(
    "Number of events",
    "Number of variables",
    "Earliest event date",
    "Latest event date",
    "Total affected population",
    "Total estimated economic loss",
    "Average severity",
    "Average response time (hours)"
  ),
  Value = c(
    nrow(data),
    ncol(data),
    format(min(data$date, na.rm = TRUE), "%Y-%m-%d"),
    format(max(data$date, na.rm = TRUE), "%Y-%m-%d"),
    comma(sum(data$affected_population, na.rm = TRUE)),
    dollar(sum(data$estimated_economic_loss_usd, na.rm = TRUE)),
    round(mean(data$severity_level, na.rm = TRUE), 2),
    round(mean(data$response_time_hours, na.rm = TRUE), 2)
  )
)
kable(overview, caption = "Dataset overview")

Dataset overview
Measure	Value
Number of events	20000
Number of variables	13
Earliest event date	2022-12-08
Latest event date	2025-12-07
Total affected population	552,824,979
Total estimated economic loss	$96,621,452,916
Average severity	5.49
Average response time (hours)	36.37

Event Frequency

Code

events_by_type <- data %>% count(disaster_type, sort = TRUE)
ggplot(events_by_type, aes(y = reorder(disaster_type, n), x = n)) +
  geom_col() + coord_flip() +
  labs(y = NULL, x = "Number of events", title = "Disaster Events by Type")

Code

knitr::kable(events_by_type, col.names = c("Disaster type", "Events"))

Disaster type	Events
Earthquake	2910
Landslide	2891
Wildfire	2870
Hurricane	2866
Drought	2863
Volcanic Eruption	2827
Flood	2773

Economic loss

Code

loss_by_type <- data %>%
  group_by(disaster_type) %>%
  summarise(
    events = n(),
    total_loss_usd = sum(estimated_economic_loss_usd, na.rm = TRUE),
    mean_loss_usd = mean(estimated_economic_loss_usd, na.rm = TRUE),
    median_loss_usd = median(estimated_economic_loss_usd, na.rm = TRUE),
    .groups = "drop"
  ) %>% arrange(desc(total_loss_usd))

ggplot(loss_by_type, aes(x = reorder(disaster_type, total_loss_usd), y = total_loss_usd)) +
  geom_col() + coord_flip() + scale_y_continuous(labels = dollar_format()) +
  labs(x = NULL, y = "Total estimated loss (USD)", title = "Estimated economic loss by disaster type")

Code

knitr::kable(loss_by_type, digits = 0)

disaster_type	events	total_loss_usd	mean_loss_usd	median_loss_usd
Earthquake	2910	14339311149	4927598	4104313
Landslide	2891	13981221492	4836120	4024796
Hurricane	2866	13916903531	4855863	4028622
Drought	2863	13831854096	4831245	4015134
Volcanic Eruption	2827	13820939149	4888907	4110298
Wildfire	2870	13602514413	4739552	3970607
Flood	2773	13128709086	4734479	3984243

Major-disaster summary

Code

major_summary <- data %>%
  group_by(is_major_disaster) %>%
  summarise(
    events = n(),
    avg_severity = mean(severity_level, na.rm = TRUE),
    avg_affected = mean(affected_population, na.rm = TRUE),
    avg_loss_usd = mean(estimated_economic_loss_usd, na.rm = TRUE),
    avg_response_hours = mean(response_time_hours, na.rm = TRUE),
    .groups = "drop"
  ) %>%
  mutate(category = if_else(is_major_disaster == 1, "Major disaster", "Other disaster")) %>%
  select(category, everything(), -is_major_disaster)
knitr::kable(major_summary, digits = 2)

category	events	avg_severity	avg_affected	avg_loss_usd	avg_response_hours
Other disaster	11999	3.49	17806.37	3107868	36.42
Major disaster	8001	8.48	42390.50	7415341	36.29

Summary by location

Code

location_summary <- data %>%
  group_by(location) %>%
  summarise(
    events = n(),
    total_affected = sum(affected_population, na.rm = TRUE),
    total_loss_usd = sum(estimated_economic_loss_usd, na.rm = TRUE),
    avg_severity = mean(severity_level, na.rm = TRUE),
    .groups = "drop"
  ) %>% arrange(desc(total_loss_usd))
knitr::kable(location_summary, digits = 2)

location	events	total_affected	total_loss_usd	avg_severity
India	2583	72183422	12652716823	5.54
Indonesia	2530	70193413	12309428718	5.51
Chile	2515	70115117	12236377078	5.56
Italy	2485	69355495	12129163356	5.48
USA	2442	67448482	11940119463	5.48
Philippines	2453	67279471	11867675791	5.46
Turkey	2515	68387863	11804815025	5.46
Japan	2477	67861716	11681156662	5.43

Outcome of Analysis from this report

Which disaster types occur most frequently?
How does the distribution of economic loss differ across disaster types?
Is higher severity associated with greater estimated economic loss?
How are response times related to the affected population?

My next Goal

Find relationships between among impact measures -

Severity level vs. Economic loss and correlation coefficient value between these two factors.
Response time vs. Affected population and correlation coefficient value between these two factors.