Population is not just about the count of people living in a given area, it’s one of the most important development indicators that influences labour markets, housing, food demand, healthcare, education, infrastructure etc. Its a key variable that influences key economic social and environmental aspects.
Forecasting is the process of making predictions about future events or trends based on historical data, current conditions, and analysis.
Forecasting population is important because it allows governments, businesses, and organizations to plan proactively rather than reactively. Since demographic changes happen slowly but have long-lasting effects, accurate forecasts provide the necessary lead time to adjust policies, infrastructure, and economies.
The data
I used the World Bank’s World Development Indicators namely: total, female and male population for every country between the years 1960–2025. That’s 66 years of history for 216 countries.
The raw download needed cleaning before it was usable:
- Regional totals were mixed in with countries. Rows like “Sub-Saharan Africa”, “High income” and “World” sit beside Kenya and Japan. Summing them would count people twice, so I removed 47 aggregates.
- Countries needed continents. The World Bank groups countries by its own regions. I built a lookup table mapping each country to its continent and UN sub-region (Kenya → Eastern Africa → Africa), which powers the continent view in the dashboard.
hOW I built the forecast
I built the pipeline in R with the fable forecasting package, in four steps.
- I had to test models to find the most accurate. I hid the last ten years of data (2016–2025) and asked three models namely : Damped-trend ETS, ARIMA and Straight-line trend to forecast them from 1960–2015 alone. Then I compared their forecasts with the actual data.
- Let the test pick the model. The test picked damped-trend ETS as the best model with an error of 1.2%(as shown in the table below), thus it was used to forecast every country to 2060, on the log scale. Modelling log population means the forecast can never go negative and treats change as a growth rate, which is how populations move.
- Keep the uncertainty. Every forecast comes with an 80% and a 95% range, not just a single number.
- Make the parts add up. I forecast each country’s total, then separately forecast the female share of it. Women = total × share, men = the rest. Women plus men always equal the total.
The back-test results, across 216 countries:
| Model | Typical error over 10 years | 95% range caught the real value | Countries where it was most accurate |
| Damped-trend ETS | 1.2% | 90% | 109 |
| ARIMA | 1.3% | 96% | 81 |
| Straight-line trend (drift) | 4.0% | 55% | 26 |
WHAT I FOUND
Here are some of the findings:
- The world’s population is projected by my model to increase from 8.19 billion in 2025 to 10.26 billion in 2060.
- Africa accounts for the largest share of that growth, adding approximately 988 million people—around 47.7% of the world’s total population increase over the period.
- Nigeria emerges as Africa’s largest population market in 2060, with approximately 398 million people in the forecast.
- Europe’s growth is much slower, with its population increasing by only about 6%, or approximately 42 million people.
And there is another interesting demographic pattern:
- Women outnumber men in most countries.In my dataset, women outnumber men in 143 of 216 countries in 2025, increasing to 145 countries by 2060. Globally, however, the balance remains close to even, at around 101 men for every 100 women.
Explore the interactive Power BI dashboard:
Why does this matter?
Because a population forecast is not just about knowing how many people there will be.
It is about asking:
- Where will they live?
- What will they need?
- Who will enter the labour market?
- Where will demand for food, housing and healthcare grow fastest?
- Which economies will have expanding consumer markets?
- And which regions will face slower or declining population growth?
These are development questions as much as they are demographic questions.
But there is an important caveat.
This is a statistical extrapolation, not a definitive demographic projection.
The model is based on historical patterns and does not explicitly account for major shocks such as wars, pandemics, migration shocks, policy changes or unexpected changes in fertility and mortality.
And one more important distinction: this analysis looks at sex composition, not age structure. Understanding the age distribution would provide an additional—and very different—picture of future demographic pressures.
If your organization has data sitting in spreadsheets that could tell a bigger story, let’s turn it into something decision-makers can actually use. Contact me
Check out more projects