The topic of this project is suicide rates worldwide, using data from the World Health Organization (WHO) on suicide statistics. This dataset includes variables such as:
Country (categorical)
Year (date/categorical)
Sex (categorical)
Age group (categorical)
Suicide rate per 100,000 population (quantitative)
Population (quantitative).
The data comes from the World Health Organization, the global public health agency of the United Nations. According to the WHO, suicide is a significant public health issue worldwide, and this dataset was compiled through national statistical agencies and health ministries submitting annual reports of deaths by suicide, aggregated and standardized by the WHO to enable international comparisons.
I chose this dataset because suicide is a critical, complex issue that affects millions of families globally. Understanding how suicide rates vary by age, sex, and country can help identify vulnerable populations and inform prevention strategies. Personally, I believe bringing attention to this data is important to reduce stigma and encourage policies to address mental health more effectively.
Load Library
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 3.5.2 ✔ tibble 3.3.0
✔ lubridate 1.9.4 ✔ tidyr 1.3.1
✔ purrr 1.0.4
── 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 errors
library(plotly)
Attaching package: 'plotly'
The following object is masked from 'package:ggplot2':
last_plot
The following object is masked from 'package:stats':
filter
The following object is masked from 'package:graphics':
layout
Rows: 43776 Columns: 6
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (3): country, sex, age
dbl (3): year, suicides_no, population
ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
Clean Dataset
# Clean, filter, and summarize suicide datasetdf_clean <- df_suicide |>select(country, year, sex, age, suicides_no, population) |>filter(year >=2008) |>mutate(rate_per_100k = suicides_no)# Summarize average suicide rate by sex and agedf_summary <- df_clean |>group_by(sex, age) |>summarise(avg_suicide_rate =mean(rate_per_100k, na.rm =TRUE) )
`summarise()` has grouped output by 'sex'. You can override using the `.groups`
argument.
# View summary table sorted by highest average ratedf_summary |>arrange(desc(avg_suicide_rate))
# A tibble: 12 × 3
# Groups: sex [2]
sex age avg_suicide_rate
<chr> <chr> <dbl>
1 male 35-54 years 608.
2 male 55-74 years 412.
3 male 25-34 years 277.
4 male 15-24 years 191.
5 female 35-54 years 167.
6 male 75+ years 156.
7 female 55-74 years 130.
8 female 25-34 years 67.1
9 female 75+ years 66.9
10 female 15-24 years 56.9
11 male 5-14 years 10.4
12 female 5-14 years 6.85
Make your 1st plot
plot3 <-ggplot(df_summary, aes(x = age, y = avg_suicide_rate, fill = sex)) +geom_col(position ="dodge") +labs(title ="Average Number of Suicides by Age Group and Sex (2008)",x ="Age Group",y ="Average Number of Suicides",fill ="Sex",caption ="Data source: WHO Suicide Statistics" ) +scale_fill_manual(values =c("male"="#0072B2", "female"="#CC79A7")) +theme_dark() +annotate("text",x =3, y =650, label ="Peak suicide rate",color ="red",size =5,fontface ="bold")plot3
When I looked at this plot, I noticed right away how middle-aged adults had the highest suicide rates, with men in that age group peaking at nearly 600 suicides on average. What really stood out to me was how big the gap is between men and women, men had way higher suicide rates in every age group. The rates were lower for younger adults and older people, but the 55-74 age group still stayed pretty high before dropping off in the oldest ages. To me, this pattern suggests that the stress and responsibilities of middle age, like work pressure or financial issues, might make things worse for people at that stage of life. I was honestly surprised by just how much higher the male rates were sometimes two or three times higher than women and it made me realize how important it is to address mental health for men, especially when they hit middle age.
Filter data for the interactive plot
country_comparison <- df_suicide |># Remove rows with missing or zero populationfilter(!is.na(suicides_no), !is.na(population), population >0) |># Calculate suicide ratemutate(suicide_rate = (suicides_no / population) *100000) |># Group by country and calculate averagegroup_by(country) |>summarise(avg_rate =mean(suicide_rate, na.rm =TRUE),.groups ='drop' ) |># Remove any countries with NA ratesfilter(!is.na(avg_rate)) |># Sort and take top 8arrange(desc(avg_rate)) |>slice_head(n =8)
I started by calculating the median average suicide rate from the country_comparison data to get a middle point that is representative of the average suicide rate between nations that is free from distortions due to high or low extremes. I then sought out the country with the mean rate closest to this median to present it as an important point of reference within the data. I used ggplot2 to plot a horizontal bar chart of the average suicide rate for every country with a color gradient to visually emphasize rate variations. To ensure the audience would notice the median, I placed an annotation above the country nearest the median value so it would be easily identifiable on the plot. Finally, I rendered the static chart interactive by converting it into a Plotly visualization so users can hover over each bar to see the comprehensive details like the country name and exact average suicide rate which makes looking for information on the plot more interesting.
Statistical Analysis - Multiple Linear Regression
# Prepare data for regressionregression_data <- df_suicide |>filter(!is.na(suicides_no), !is.na(population), population >0) |>mutate(suicide_rate = (suicides_no / population) *100000,male =ifelse(sex =="male", 1, 0),age_numeric =case_when( age =="5-14 years"~1, age =="15-24 years"~2, age =="25-34 years"~3, age =="35-54 years"~4, age =="55-74 years"~5, age =="75+ years"~6 ) ## chatgpt assistance in making this code ) |>filter(!is.na(suicide_rate), !is.na(age_numeric), suicide_rate <200)
# Build modelsuicide_model <-lm(suicide_rate ~ male + age_numeric + year, data = regression_data)summary(suicide_model)
Call:
lm(formula = suicide_rate ~ male + age_numeric + year, data = regression_data)
Residuals:
Min 1Q Median 3Q Max
-33.684 -9.419 -1.524 5.194 164.202
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 253.573176 17.229686 14.72 <2e-16 ***
male 14.792772 0.175172 84.45 <2e-16 ***
age_numeric 4.212839 0.051293 82.13 <2e-16 ***
year -0.131359 0.008619 -15.24 <2e-16 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 16.63 on 36047 degrees of freedom
Multiple R-squared: 0.2812, Adjusted R-squared: 0.2812
F-statistic: 4701 on 3 and 36047 DF, p-value: < 2.2e-16
The diagnostic plots for the regression model help us assess how well the model fits the data and whether the assumptions of linear regression are met.They show things like the distribution of residuals, the relationship between fitted values and residuals, and any potential outliers or influential points. From these plots, I can see if there are any issues with non-linearity or unusual data points that might affect the reliability of the model’s results.
Research
For my research, I decided to focus on why men have such a higher rate of suicide compared to women, which was clear in my first plot. According to the American Institute for Boys and Men, in 2023, 39,045 men and 10,270 women took their own lives, demonstrating that the suicide risk for men is nearly four times higher than for women (American Institute for Boys and Men, 2023). Our World in Data explains that while the exact reasons for this gender gap are still debated by researchers, several factors likely contribute to this disparity, including differences in the lethality of suicide methods chosen, stigma around seeking help, varying social pressures between genders, and higher rates of alcohol and drug abuse among men (Our World in Data, 2024). The data from Our World in Data also reveals that this pattern is consistent across most countries globally, not just in the United States (Our World in Data, 2024). Understanding these complex factors helped me realize how important it is to address the multiple barriers that prevent men from seeking mental health support and to develop targeted prevention strategies that acknowledge these gender-specific risk factors.
Sources:
“Male Suicide: Patterns and Recent Trends.” American Institute for Boys and Men, 6 June 2025, aibm.org/research/male-suicide/.
Ritchie, Hannah. “Suicide Rates Are Higher in Men than Women.” Our World in Data, ourworldindata.org/data-insights/suicide-rates-are-higher-in-men-than-women. Accessed 4 July 2025.
Reflection
Looking back on this project, I’m really proud of what I was able to put together, but there were definitely a few things I wish I could have done better or included. One thing I wanted to show was a time series plot to explore how suicide rates changed over time for each country or age group, but I couldn’t get it to work properly with my data because there were too many missing values across different years. When I got to the linear regression section of my project, I found it pretty confusing at first. I wasn’t sure how to set up the model properly with categorical variables like sex and age group, or how to interpret things like the coefficients and p-values. I ended up spending a lot of time researching online and trying different combinations of variables, but I still couldn’t figure everything out on my own.Even after finishing the linear regression part of my project, I still don’t feel completely confident on how it turned out. Another issue I had when making this was the inclusion of caption for my plotly, I tried adding it to my labs multiple times but even though the code runs it never shows up on the actual plot.
