A Quick Look Into Some Suicide Statistics

UCLA DataRes
May 28 · 6 min read

By Chingyi Ie and Richard Yim

Overview

In order to figure out some trends in global suicide rates, we looked at the “Suicide Rates Overview 1985 to 2016” dataset from Kaggle.com. We then took a look into suicide trends in the US and how they correlated with some economic trends over the same time period.

Suicide Dataset Trends/Insights

We first decided to take a quick look at the total suicides for each year. However, since some countries have missing data for some years, it would be difficult and inaccurate to compare the progression of the world’s total suicide rate over the years, as the lack of data for some countries would result in a lower value for some years than we would expect. Hence, we decided to take a look at the suicide rates in the United States alone.

Fig 1: Total Suicide Numbers Over the Years in United States

As we see, total suicides in the United States seem to be relatively constant over the years 1985 to 2002, before a steady increase from 2003 to 2014. The figure above also shows a huge gap in suicide rates for male and females:male suicides seemed to outnumber female suicides..

This prompted us to further investigate the relation of sex and suicide rates. To do this, we decided to examine suicide by age group. We first did an overall examination of total suicide by age group.

Figure 2: Total Suicides per Age group

Here, we are using the total data for the world, and while some countries contain missing data for total suicides, there is no missing data based on age group or sex, so the comparison will not be inaccurate. In Figure 2, we see that total suicides are highest for ages 35–54, and lowest for ages 5–14.

We then did an examination of male and female suicide rates per age group. To focus purely on the rate of suicide and not the overall numbers, we decided to normalize the data, or convert suicide rates, to decimal values between 0 and 1, for suicide trend visualization

Figure 3: Male vs Female Suicide Rates for ages 5–14 years
Figure 4: Male vs Female Suicide Rates for ages 25–34 years
Figure 6: Male vs Female Suicide Rates for ages 35–54 years

Through this investigation we found an interesting yet disturbing trend, which is that in recent years, the change in the rate of suicide for females is actually higher than that of males for people below the age of 55 years! A look at figures 3,4,5, and 6 show that the change in female suicide rates surpassed that of males beginning as early as around 2003 for women aged 35–54 and as late as around 2008 for 25–34 (though 2008 was still 11 years ago!).

A quick look at the data for the United States also shows a similar rising trend for female suicides, especially in females from ages 5 to 34 (which are the youngest 3 age groups). Those 3 age groups showed an increasing suicide rate which eventually met or surpassed the male suicide rate for those age groups, We see an example of this in figure 7 below.

Figure 7

Another interesting trend we found is that females in the United States above the age of 75 have way higher suicide rates than males until very recently

Figure 8: Male vs Female Suicide Rates in the US for ages 35–54 years

From figure 8, we see that the first jump in suicide rates occurred around 1990, and the huge gap in suicide rates continued all the way until 2013. Hence we see that although overall more males have committed suicide than females, when looking at the change in rates itself, there are more females committing suicide on a year over year basis than males.


We now take a look at year over year suicide rates in the United States per gender as well as factors like unemployment and K-12 enrollment rates.

Figure 9: Total, Male, Female, GDP per Capita

We observe an interesting increase of suicide rates year over year starting in 2000. We can see that although the population has been increasing since the late 1900s, there is an obvious increase in suicides in the US beginning in 1999. Also, note that regardless of the increase in GDP per capita, suicides have been steadily increasing at a pace greater than population growth.

Figure 10: Total Revenue, Expenditure, Instructional Expenditure
Figure 11: Mean, Median, and Standard Deviation of Unemployment Rates by County

In addition to Figure 9, we have in Figure 10 and 11 showing year over year spending in education by the government and the average and median county unemployment rate. , respectively. We can see a base support of the county unemployment rate increasing from 2000 to 2015, with a mostly small standard deviation from the mean. In addition we can see an increase in total expenditures by the government matching the increasing unemployment rate.

Figure 12: Population Suicide, Population Enrolled, Median County Unemployment

In this final chart we combine the overall trends of population suicide rates, population enrollment rates, as well the median county unemployment rate. Interestingly enough, we see that population suicide rates seem to correlated with increases in the median county unemployment rate. Observing the the lowest bounding rate from 2000 to 2015, we see a correlation between the median county unemployment rate and the suicide rate, as both increase over this time period. The plot was generated by scaling the enrollment rate by a factor of 1000. The median county unemployment rate was scaled down by 50,000 and pushed up by 0.0003. The purpose was to simply visualize the trend and relationships between unemployment, suicide, and enrollment rates in the US year over year from 1993 to 2015.

Conclusion

Globally we are seeing a slight decrease in suicides, yet we see a general increase in suicide rates across the board from 1993 to 2015 in the United States. Moreover, the dominance of suicides per gender has been overtaken mostly by females across all age groups starting around 2005; however, for suicides over the age of 75, males make up the majority of the victims. Finally, we see that with the baseline rise in median county unemployment in the US the rate at which the percentage of the population commits suicide year over year has increased as well.

Logistically, there are gaps across the board for all datasets. Aside from the gaps of data, we wished there would’ve been some attributes left out and others left it. Attributes such as the World Health Organisation’s suicide statistics inclusion of categorical information of a generation; arbitrary expenditure information such as “other expenditures, local revenues” in the education dataset, were seen as redundant and/or meaningless in the context of our study. Furthermore, the suicide dataset ideally would have been more descriptive with concern to suicide statistics per state, and the unemployment data set could have included state overall unemployment rates by state and the the country overall. In future projects we wish to acquire more advanced statistical techniques and web scraping skills to have better data analysis as well as acquire more and higher quality data, though this is a more long term goal.

Sources for datasets:

https://www.kaggle.com/jayrav13/unemployment-by-county-us

https://www.kaggle.com/szamil/who-suicide-statistics

https://www.kaggle.com/noriuk/us-education-datasets-unification-project

UCLA DataRes

Written by

UCLA’s first Data Science Club. datares.github.io/, facebook.com/ucladatares, linkedin.com/company/ucla-datares

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade