Comprehensive Disaster Impact Analysis: Patterns, Vulnerabilities and Trends in Asia (2000-2025)
Swaroop
2025-11-09
Introduction
Natural and technological disasters pose significant threats to human life, economic stability, and sustainable development across the globe. Asia, being the world’s largest and most populous continent, experiences a disproportionately high share of global disasters, making it a critical region for disaster risk analysis. This comprehensive study analyzes disaster patterns, impacts, and trends across Asian countries using the Emergency Events Database (EM-DAT) maintained by the Centre for Research on the Epidemiology of Disasters (CRED).
Dataset Overview
The EM-DAT database is internationally recognized as the most comprehensive global disaster database, systematically collecting and validating information on natural and technological disasters since 1900. For this analysis, we focus on the Asian subset of the EM-DAT database, covering the period from 2000 to 2025. This dataset encompasses:
- Temporal Coverage: 25 years of disaster records (January 2000 - April 2025)
- Geographic Scope: All Asian countries including East Asia, South Asia, Southeast Asia, Central Asia, and Western Asia
- Disaster Categories: Both natural disasters (geophysical, meteorological, hydrological, climatological, biological) and technological disasters (industrial accidents, transport accidents, miscellaneous accidents)
- Total Records: Over 2,500 disaster events specific to the Asian region
Key Variables in the Dataset
The dataset contains over 40 variables capturing multiple dimensions of disaster impacts:
- Identification Variables: Disaster number, country, ISO codes, region, subregion
- Temporal Variables: Start date, end date, duration of disaster
- Classification Variables: Disaster group, subgroup, type, subtype
- Human Impact Variables: Total deaths, number injured, number affected, number homeless
- Economic Impact Variables: Total damage, reconstruction costs, insured damage (both nominal and CPI-adjusted values)
- Physical Characteristics: Magnitude, magnitude scale, geographic coordinates (latitude/longitude)
- Response Variables: Aid contributions, OFDA/BHA response, appeals, declarations
Structure of the Report
Following this introduction, the report is organized into five main sections:
- Objectives: Detailed listing of 20 specific analytical objectives
- Preliminary Analysis: Data cleaning, preprocessing, and quality assessment
- Detailed Analysis: Comprehensive statistical analysis and visualization of disaster patterns
- Outcomes: Summary of achieved objectives and key findings
- Results and Discussions: Interpretation of findings and their implications for disaster risk management
This analysis provides critical insights for policymakers, disaster management agencies, and international organizations working toward disaster risk reduction and resilience building in Asia.
Objective
The objectives planned for the mini project are as below:
- To analyze temporal trends of disaster frequency across 25 years (2000-2025) and identify significant pattern changes
- To analyze seasonal and monthly distribution of disasters to identify peak disaster periods globally
- To classify disasters into major categories and subcategories (Natural, Technological, Meteorological, Hydrological, etc.) for comprehensive analytical understanding
- To analyze the yearly and seasonal distribution of disasters and identify significant trends or pattern changes in frequency over the 25-year period
- To investigate the regional and country-wise distribution of disasters to identify the most disaster-prone regions and top 10 affected countries globally
- To analyze the impact of disasters on human life by studying total deaths, number of injured, affected, and homeless populations across different disaster types
- To assess the economic impact of disasters by evaluating total damage, reconstruction costs, and insured losses with CPI adjustments
- To study the relationship between disaster magnitude and its human/economic consequences using correlation analysis and statistical inference
- To identify the top 10 deadliest disasters and top 10 most economically damaging disasters in the dataset
- To visualize the frequency and intensity of different disaster types
using bar charts, boxplots, histograms, and density plots in
ggplot2 - To create geographic visualizations mapping global disaster hotspots using latitude and longitude coordinates
- To develop heat maps showing disaster intensity across regions and years for pattern identification
- To compare disaster severity and impacts between developed and developing nations using descriptive and graphical methods
- To analyze decade-wise changes in disaster patterns (2000-2010 vs 2011-2020 vs 2021-2025) to identify evolving trends
- To explore the duration of disasters by calculating start and end dates to identify long-lasting versus short-lived events
- To detect potential outliers or extreme cases in deaths, damages, or magnitude values using boxplot and statistical analysis
- To examine missing data patterns across all variables and assess their impact on analysis quality
- To create faceted visualizations displaying distribution of disaster types across continents for comparative analysis
- To develop a comprehensive dashboard integrating multiple metrics (deaths, affected population, economic losses) for top 20 disaster-prone countries
- To generate integrated visualization showing top 20 most disaster-prone countries with multiple impact metrics
Preliminary Analysis
Data Import and Structure Verification
# Load required libraries
library(tidyverse)## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.1.4 ✔ readr 2.1.5
## ✔ forcats 1.0.0 ✔ stringr 1.5.1
## ✔ ggplot2 4.0.1 ✔ tibble 3.3.0
## ✔ lubridate 1.9.4 ✔ tidyr 1.3.1
## ✔ purrr 1.1.0
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errorsdata <- read.csv("extracted_data/Asia.csv", stringsAsFactors = FALSE)
disaster_data <- data
# Display basic information
cat("Dataset Dimensions:", nrow(disaster_data), "rows and", ncol(disaster_data), "columns\n")## Dataset Dimensions: 6531 rows and 46 columnscat("Number of Countries:", length(unique(disaster_data$Country)), "\n")## Number of Countries: 50cat("Number of unique disaster types:", length(unique(disaster_data$`Disaster Type`)), "\n\n")## Number of unique disaster types: 0# Check column names
cat("Column names in dataset:\n")## Column names in dataset:names(disaster_data)## [1] "DisNo."
## [2] "Historic"
## [3] "Classification.Key"
## [4] "Disaster.Group"
## [5] "Disaster.Subgroup"
## [6] "Disaster.Type"
## [7] "Disaster.Subtype"
## [8] "External.IDs"
## [9] "Event.Name"
## [10] "ISO"
## [11] "Country"
## [12] "Subregion"
## [13] "Region"
## [14] "Location"
## [15] "Origin"
## [16] "Associated.Types"
## [17] "OFDA.BHA.Response"
## [18] "Appeal"
## [19] "Declaration"
## [20] "AID.Contribution...000.US.."
## [21] "Magnitude"
## [22] "Magnitude.Scale"
## [23] "Latitude"
## [24] "Longitude"
## [25] "River.Basin"
## [26] "Start.Year"
## [27] "Start.Month"
## [28] "Start.Day"
## [29] "End.Year"
## [30] "End.Month"
## [31] "End.Day"
## [32] "Total.Deaths"
## [33] "No..Injured"
## [34] "No..Affected"
## [35] "No..Homeless"
## [36] "Total.Affected"
## [37] "Reconstruction.Costs...000.US.."
## [38] "Reconstruction.Costs..Adjusted...000.US.."
## [39] "Insured.Damage...000.US.."
## [40] "Insured.Damage..Adjusted...000.US.."
## [41] "Total.Damage...000.US.."
## [42] "Total.Damage..Adjusted...000.US.."
## [43] "CPI"
## [44] "Admin.Units"
## [45] "Entry.Date"
## [46] "Last.Update"Data Cleaning and Preprocessing
library(tidyverse)
library(lubridate)
# Convert all empty strings to NA
disaster_data[disaster_data == ""] <- NA
disaster_clean <- disaster_data %>%
mutate(
# --- SAFE DATE CONSTRUCTION ---
Start_Date = make_date(
year = as.integer(Start.Year),
month = as.integer(Start.Month),
day = as.integer(Start.Day)
),
End_Date = make_date(
year = as.integer(End.Year),
month = as.integer(End.Month),
day = as.integer(End.Day)
),
Duration_Days = as.numeric(End_Date - Start_Date),
# --- TEMPORAL VARIABLES ---
Year = as.integer(Start.Year),
Month = as.integer(Start.Month),
Season = case_when(
Month %in% c(12,1,2) ~ "Winter",
Month %in% c(3,4,5) ~ "Spring",
Month %in% c(6,7,8) ~ "Summer",
Month %in% c(9,10,11) ~ "Fall",
TRUE ~ NA_character_
),
Decade = case_when(
Year >= 2000 & Year <= 2010 ~ "2000-2010",
Year >= 2011 & Year <= 2020 ~ "2011-2020",
Year >= 2021 & Year <= 2025 ~ "2021-2025",
TRUE ~ NA_character_
),
Duration_Category = case_when(
is.na(Duration_Days) ~ "Unknown",
Duration_Days == 0 ~ "Single Day",
Duration_Days <= 7 ~ "Short (1-7 days)",
Duration_Days <= 30 ~ "Medium (8-30 days)",
Duration_Days <= 90 ~ "Long (31-90 days)",
Duration_Days > 90 ~ "Very Long (>90 days)"
),
# --- NUMERIC CLEANING ---
Total_Deaths = as.numeric(gsub(",", "", Total.Deaths)),
Injured = as.numeric(gsub(",", "", No..Injured)),
Affected = as.numeric(gsub(",", "", No..Affected)),
Homeless = as.numeric(gsub(",", "", No..Homeless)),
Total_Affected = as.numeric(gsub(",", "", Total.Affected)),
Total_Damage = as.numeric(gsub(",", "", Total.Damage...000.US..)),
Total_Damage_Adjusted = as.numeric(gsub(",", "", Total.Damage..Adjusted...000.US..)),
Reconstruction_Costs = as.numeric(gsub(",", "", Reconstruction.Costs...000.US..)),
Insured_Damage = as.numeric(gsub(",", "", Insured.Damage...000.US..)),
# --- MAGNITUDE ---
Magnitude_Value = suppressWarnings(as.numeric(gsub("[^0-9\\.]", "", Magnitude))),
# --- GEO COORDINATES ---
Lat = as.numeric(Latitude),
Lon = as.numeric(Longitude),
# --- KEEP SAME NAMES AS YOUR ORIGINAL PIPELINE EXPECTS ---
Disaster_Group = Disaster.Group,
Disaster_Type = Disaster.Type,
Disaster_Subtype = Disaster.Subtype
)
cat("Data cleaning completed successfully.\n")## Data cleaning completed successfully.cat("Rows:", nrow(disaster_clean), "\n")## Rows: 6531cat("Columns:", ncol(disaster_clean), "\n")## Columns: 69Detailed Analysis
1. Temporal Trends Analysis (Objectives 1, 2, 4)
# Yearly trends
yearly_trends <- disaster_clean %>%
group_by(Year) %>%
summarise(
Count = n(),
Deaths = sum(Total_Deaths, na.rm = TRUE),
Affected = sum(Total_Affected, na.rm = TRUE),
.groups = 'drop'
)
# Trend model
trend_model <- lm(Count ~ Year, data = yearly_trends)
cat("Annual Trend Analysis:\n")## Annual Trend Analysis:cat("Average increase per year:", round(coef(trend_model)[2], 2), "disasters\n")## Average increase per year: -6.87 disasterscat("R-squared:", round(summary(trend_model)$r.squared, 3), "\n\n")## R-squared: 0.634# Monthly and seasonal patterns
monthly_pattern <- disaster_clean %>%
group_by(Month) %>%
summarise(Count = n(), .groups = 'drop')
seasonal_pattern <- disaster_clean %>%
count(Season)
# Visualizations
p1 <- ggplot(yearly_trends, aes(x = Year, y = Count)) +
geom_line(color = "darkblue", size = 1) +
geom_point(color = "red") +
geom_smooth(method = "lm", se = TRUE, alpha = 0.2) +
labs(title = "Yearly Disaster Frequency Trend", x = "Year", y = "Number of Disasters") +
theme_minimal()## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` instead.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.p2 <- ggplot(monthly_pattern, aes(x = factor(Month), y = Count)) +
geom_bar(stat = "identity", fill = "steelblue") +
labs(title = "Monthly Distribution", x = "Month", y = "Frequency") +
scale_x_discrete(labels = month.abb) +
theme_minimal()
p3 <- ggplot(seasonal_pattern, aes(x = Season, y = n, fill = Season)) +
geom_bar(stat = "identity") +
labs(title = "Seasonal Pattern", x = "Season", y = "Frequency") +
theme_minimal() +
theme(legend.position = "none")
print(p1)## `geom_smooth()` using formula = 'y ~ x'
print(p2)
print(p3)
### 2. Disaster Classification (Objective 3)
# Classification summary
classification_summary <- disaster_clean %>%
group_by(Disaster_Group, Disaster_Type) %>%
summarise(
Count = n(),
Deaths = sum(Total_Deaths, na.rm = TRUE),
Economic_Loss = sum(Total_Damage, na.rm = TRUE),
.groups = 'drop'
) %>%
arrange(desc(Count))
# Top disaster types
top_types <- classification_summary %>%
group_by(Disaster_Type) %>%
summarise(Total_Count = sum(Count), .groups = 'drop') %>%
arrange(desc(Total_Count)) %>%
head(10)
# Visualization
ggplot(top_types, aes(x = reorder(Disaster_Type, Total_Count), y = Total_Count)) +
geom_bar(stat = "identity", fill = "darkgreen") +
coord_flip() +
labs(title = "Top 10 Disaster Types in Asia", x = "Disaster Type", y = "Frequency") +
theme_minimal()
cat("\nDisaster Classification Summary:\n")##
## Disaster Classification Summary:cat("Natural Disasters:", sum(disaster_clean$Disaster_Group == "Natural", na.rm = TRUE), "\n")## Natural Disasters: 4015cat("Technological Disasters:", sum(disaster_clean$Disaster_Group == "Technological", na.rm = TRUE), "\n")## Technological Disasters: 25163. Geographic Distribution (Objectives 5,11)
# Country analysis
country_impact <- disaster_clean %>%
group_by(Country) %>%
summarise(
Disasters = n(),
Deaths = sum(Total_Deaths, na.rm = TRUE),
Affected = sum(Total_Affected, na.rm = TRUE),
Economic_Loss = sum(Total_Damage, na.rm = TRUE),
Avg_Lat = mean(Lat, na.rm = TRUE),
Avg_Lon = mean(Lon, na.rm = TRUE),
.groups = 'drop'
) %>%
arrange(desc(Disasters))
top10_countries <- head(country_impact, 10)
# Bar chart for top 10 countries
ggplot(top10_countries, aes(x = reorder(Country, Disasters), y = Disasters)) +
geom_bar(stat = "identity", fill = "darkred") +
coord_flip() +
labs(title = "Top 10 Disaster-Prone Countries", x = "", y = "Number of Disasters") +
theme_minimal()
print("Top 5 Most Affected Countries:")## [1] "Top 5 Most Affected Countries:"print(top10_countries[1:5, c("Country", "Disasters", "Deaths")])## # A tibble: 5 × 3
## Country Disasters Deaths
## <chr> <int> <dbl>
## 1 China 1355 134151
## 2 India 820 104492
## 3 Indonesia 564 196463
## 4 Philippines 474 30528
## 5 Pakistan 321 93464# LEAFLET MAP 1: Disaster Hotspots by Location
library(leaflet) # Ensure leaflet is loaded
map_data <- disaster_clean %>%
filter(!is.na(Lat) & !is.na(Lon))
hotspot_map <- leaflet(map_data) %>%
addTiles() %>%
setView(lng = 100, lat = 20, zoom = 3) %>% # Center on Asia
addCircleMarkers(
~Lon, ~Lat,
radius = ~ifelse(Total_Deaths > 0, log(Total_Deaths + 1), 1),
color = ~ifelse(Disaster_Group == "Natural", "red", "blue"),
fillOpacity = 0.5,
stroke = TRUE,
weight = 1,
popup = ~paste(
"<b>Location:</b>", Location, "<br>",
"<b>Country:</b>", Country, "<br>",
"<b>Type:</b>", Disaster_Type, "<br>",
"<b>Year:</b>", Year, "<br>",
"<b>Deaths:</b>", Total_Deaths, "<br>",
"<b>Affected:</b>", Total_Affected
),
clusterOptions = markerClusterOptions()
) %>%
leaflet::addLegend(
position = "bottomright",
colors = c("red", "blue"),
labels = c("Natural Disasters", "Technological Disasters"),
title = "Disaster Type"
)
# Display the hotspot map
hotspot_map# LEAFLET MAP 2: Country-level Summary Map
country_map_data <- country_impact %>%
filter(!is.na(Avg_Lat) & !is.na(Avg_Lon))
country_map <- leaflet(country_map_data) %>%
addTiles() %>%
setView(lng = 100, lat = 20, zoom = 3) %>%
addCircles(
~Avg_Lon, ~Avg_Lat,
radius = ~sqrt(Disasters) * 30000, # Scale circle size by disasters
color = "orange",
fillColor = "yellow",
fillOpacity = 0.6,
popup = ~paste(
"<b>Country:</b>", Country, "<br>",
"<b>Total Disasters:</b>", Disasters, "<br>",
"<b>Total Deaths:</b>", format(Deaths, big.mark = ","), "<br>",
"<b>Total Affected:</b>", format(Affected, big.mark = ","), "<br>",
"<b>Economic Loss:</b> $", format(Economic_Loss/1000, big.mark = ","), "M"
),
label = ~Country
)
# Display country map
country_map# LEAFLET MAP 3: Simple point map as alternative
point_map <- leaflet(map_data) %>%
addTiles() %>%
setView(lng = 100, lat = 20, zoom = 3) %>%
addMarkers(
~Lon, ~Lat,
popup = ~paste(Location, "-", Disaster_Type, "-", Year),
clusterOptions = markerClusterOptions()
)
# Display point map
point_map4. Human Impact Analysis (Objective 6)
human_impact <- disaster_clean %>%
group_by(Disaster_Type) %>%
summarise(
Total_Deaths = sum(Total_Deaths, na.rm = TRUE),
Total_Injured = sum(Injured, na.rm = TRUE),
Total_Affected = sum(Affected, na.rm = TRUE),
.groups = 'drop'
) %>%
arrange(desc(Total_Deaths)) %>%
head(5)
# Visualization
human_impact %>%
pivot_longer(cols = -Disaster_Type, names_to = "Impact", values_to = "Count") %>%
ggplot(aes(x = Disaster_Type, y = Count, fill = Impact)) +
geom_bar(stat = "identity", position = "dodge") +
scale_y_log10() +
labs(title = "Human Impact by Disaster Type") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
5. Economic Impact Analysis (Objective 7)
economic_impact <- disaster_clean %>%
filter(!is.na(Total_Damage)) %>%
group_by(Year) %>%
summarise(
Nominal = sum(Total_Damage, na.rm = TRUE),
Adjusted = sum(Total_Damage_Adjusted, na.rm = TRUE),
.groups = 'drop'
)
ggplot(economic_impact, aes(x = Year)) +
geom_line(aes(y = Nominal, color = "Nominal"), size = 1) +
geom_line(aes(y = Adjusted, color = "CPI-Adjusted"), size = 1) +
labs(title = "Economic Impact Over Time", y = "Damage ('000 USD)") +
theme_minimal()
6. Correlation Analysis (Objective 8)
correlation_data <- disaster_clean %>%
filter(!is.na(Magnitude_Value) & !is.na(Total_Deaths))
cor_value <- cor(correlation_data$Magnitude_Value,
correlation_data$Total_Deaths,
use = "complete.obs")
cat("Correlation between Magnitude and Deaths:", round(cor_value, 3), "\n")## Correlation between Magnitude and Deaths: -0.02ggplot(correlation_data, aes(x = Magnitude_Value, y = log(Total_Deaths + 1))) +
geom_point(alpha = 0.5) +
geom_smooth(method = "lm", se = TRUE) +
labs(title = "Magnitude vs Deaths Relationship") +
theme_minimal()## `geom_smooth()` using formula = 'y ~ x'
### 7. Top Disasters (Objective 9)
# Top 10 deadliest
top_deadly <- disaster_clean %>%
arrange(desc(Total_Deaths)) %>%
select(Year, Country, Disaster_Type, Total_Deaths) %>%
head(10)
# Top 10 economic
top_economic <- disaster_clean %>%
arrange(desc(Total_Damage)) %>%
select(Year, Country, Disaster_Type, Total_Damage) %>%
head(10)
print("Top 10 Deadliest Disasters:")## [1] "Top 10 Deadliest Disasters:"print(top_deadly)## Year Country Disaster_Type Total_Deaths
## 1 2004 Indonesia Earthquake 165708
## 2 2008 Myanmar Storm 138366
## 3 2008 China Earthquake 87476
## 4 2005 Pakistan Earthquake 73338
## 5 2023 Türkiye Earthquake 53000
## 6 2004 Sri Lanka Earthquake 35399
## 7 2003 Iran (Islamic Republic of) Earthquake 26796
## 8 2001 India Earthquake 20005
## 9 2011 Japan Earthquake 19846
## 10 2004 India Earthquake 163898. Comprehensive Visualizations (Objective 10)
# Boxplot
disaster_clean %>%
filter(Total_Deaths > 0) %>%
ggplot(aes(x = Disaster_Group, y = log(Total_Deaths), fill = Disaster_Group)) +
geom_boxplot() +
labs(title = "Death Distribution by Group") +
theme_minimal()
# Histogram
disaster_clean %>%
filter(!is.na(Total_Damage) & Total_Damage > 0) %>%
ggplot(aes(x = log(Total_Damage))) +
geom_histogram(bins = 30, fill = "darkgreen", alpha = 0.7) +
labs(title = "Economic Damage Distribution") +
theme_minimal()
# Density
disaster_clean %>%
filter(!is.na(Total_Affected) & Total_Affected > 0) %>%
ggplot(aes(x = log(Total_Affected))) +
geom_density(fill = "purple", alpha = 0.5) +
labs(title = "Affected Population Density") +
theme_minimal()
### 9. Interactive Maps (Objectives 11, 12)
# Prepare map data
map_data <- disaster_clean %>%
filter(!is.na(Lat) & !is.na(Lon))
# Create interactive disaster map
disaster_map <- leaflet(map_data) %>%
addTiles() %>%
setView(lng = 100, lat = 20, zoom = 3) %>%
addCircleMarkers(
~Lon, ~Lat,
radius = ~log(Total_Deaths + 1),
color = ~ifelse(Disaster_Group == "Natural", "red", "blue"),
fillOpacity = 0.5,
popup = ~paste(
"Location:", Location, "<br>",
"Type:", Disaster_Type, "<br>",
"Deaths:", Total_Deaths
),
clusterOptions = markerClusterOptions()
) %>%
leaflet::addLegend(
colors = c("red", "blue"),
labels = c("Natural", "Technological"),
title = "Disaster Type"
)
disaster_map# Create heatmap using ggplot2 (since heatmap function might cause issues)
heatmap_data <- disaster_clean %>%
group_by(Subregion, Year) %>%
summarise(Count = n(), .groups = 'drop')
ggplot(heatmap_data %>% filter(Year >= 2010),
aes(x = Year, y = Subregion, fill = Count)) +
geom_tile() +
scale_fill_gradient2(low = "white", mid = "orange", high = "darkred") +
labs(title = "Disaster Intensity Heatmap") +
theme_minimal()
10. Development Comparison (Objective 13)
developed <- c("Japan", "South Korea", "Singapore", "Israel", "UAE", "Qatar")
development <- disaster_clean %>%
mutate(Status = ifelse(Country %in% developed, "Developed", "Developing")) %>%
group_by(Status) %>%
summarise(
Disasters = n(),
Deaths = sum(Total_Deaths, na.rm = TRUE),
.groups = 'drop'
)
ggplot(development, aes(x = Status, y = Disasters, fill = Status)) +
geom_bar(stat = "identity") +
labs(title = "Developed vs Developing Nations") +
theme_minimal()
### 11. Decade Analysis (Objective 14)
decade_analysis <- disaster_clean %>%
group_by(Decade) %>%
summarise(
Disasters = n(),
Deaths = sum(Total_Deaths, na.rm = TRUE),
.groups = 'drop'
)
ggplot(decade_analysis, aes(x = Decade, y = Disasters, fill = Decade)) +
geom_bar(stat = "identity") +
labs(title = "Decade-wise Analysis") +
theme_minimal()
### 12. Duration Analysis (Objective 15)
duration <- disaster_clean %>%
filter(!is.na(Duration_Days)) %>%
mutate(
Duration_Category = case_when(
Duration_Days == 0 ~ "Single Day",
Duration_Days <= 7 ~ "Short",
Duration_Days <= 30 ~ "Medium",
TRUE ~ "Long"
)
)
ggplot(duration, aes(x = Duration_Category)) +
geom_bar(fill = "steelblue") +
labs(title = "Disaster Duration Distribution") +
theme_minimal()
13. Outlier Detection (Objective 16)
# Identify outliers
Q1 <- quantile(disaster_clean$Total_Deaths, 0.25, na.rm = TRUE)
Q3 <- quantile(disaster_clean$Total_Deaths, 0.75, na.rm = TRUE)
IQR <- Q3 - Q1
outliers <- disaster_clean %>%
filter(Total_Deaths > Q3 + 1.5 * IQR) %>%
arrange(desc(Total_Deaths)) %>%
head(10)
print("Top Outliers:")## [1] "Top Outliers:"print(outliers[, c("Year", "Country", "Disaster_Type", "Total_Deaths")])## Year Country Disaster_Type Total_Deaths
## 1 2004 Indonesia Earthquake 165708
## 2 2008 Myanmar Storm 138366
## 3 2008 China Earthquake 87476
## 4 2005 Pakistan Earthquake 73338
## 5 2023 Türkiye Earthquake 53000
## 6 2004 Sri Lanka Earthquake 35399
## 7 2003 Iran (Islamic Republic of) Earthquake 26796
## 8 2001 India Earthquake 20005
## 9 2011 Japan Earthquake 19846
## 10 2004 India Earthquake 1638914. Missing Data Analysis (Objective 17)
missing <- disaster_clean %>%
summarise_all(~sum(is.na(.))) %>%
pivot_longer(everything(), names_to = "Variable", values_to = "Missing") %>%
mutate(Percentage = Missing / nrow(disaster_clean) * 100) %>%
filter(Percentage > 0) %>%
arrange(desc(Percentage))
ggplot(missing %>% head(10),
aes(x = reorder(Variable, Percentage), y = Percentage)) +
geom_bar(stat = "identity", fill = "coral") +
coord_flip() +
labs(title = "Missing Data Patterns") +
theme_minimal()
15. Faceted Visualization (Objective 18)
facet_data <- disaster_clean %>%
filter(Disaster_Type %in% c("Flood", "Storm", "Earthquake", "Drought")) %>%
count(Subregion, Disaster_Type)
ggplot(facet_data, aes(x = Disaster_Type, y = n, fill = Disaster_Type)) +
geom_bar(stat = "identity") +
facet_wrap(~Subregion) +
labs(title = "Disaster Types by Subregion") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
### 16. Dashboard for Top 20 Countries (Objectives 19, 20)
top20 <- disaster_clean %>%
group_by(Country) %>%
summarise(
Disasters = n(),
Deaths = sum(Total_Deaths, na.rm = TRUE),
Affected = sum(Total_Affected, na.rm = TRUE),
Economic_Loss = sum(Total_Damage, na.rm = TRUE),
.groups = 'drop'
) %>%
arrange(desc(Disasters)) %>%
head(20)
# Multi-metric visualization
ggplot(top20 %>% head(10),
aes(x = log(Deaths + 1), y = log(Economic_Loss + 1),
size = Affected/1e6, color = Disasters)) +
geom_point(alpha = 0.7) +
geom_text(aes(label = Country), size = 3, check_overlap = TRUE) +
scale_color_gradient(low = "yellow", high = "red") +
labs(title = "Integrated Impact Analysis",
size = "Affected (M)", color = "Disasters") +
theme_minimal()