# Births and population around the world

## Findings
Across the 74 annual UN estimates from 1950 through 2023, world births reached their maximum in 2012 at 146,054,881. In 2023 the estimate was 132,110,264, a decline of 9.5475186481% (displayed as 9.5%, rounded once to one decimal).

World population on 1 July rose from 7,201,202,484 in 2012 to 8,091,734,930 in 2023, an increase of 12.3664408545% (displayed as 12.4%). These are modeled historical estimates, not exact census counts. This chart makes no claim about what happens after 2023.

## Files
- peak_births.png: 1600 × 1800 pixel chart, two aligned line panels with zero baselines.
- world_births_population_1950_2023.csv: all 74 world annual rows used in the chart.
- country_births_population_1950_2023.csv: 17,538 annual rows covering all 237 countries/areas in the source, with no overlapping regional aggregates.
- render.py: reproducible Python 3 renderer and data extraction. Dependencies: pandas and Pillow. Run `python render.py` in a writable folder with these installed. Fetches the source snapshot when absent, checks its SHA-256, validates the chart series and exports the files.
- validation.json: source identity and exact percentage changes.

## Source and reuse
United Nations, Department of Economic and Social Affairs, Population Division (2024). World Population Prospects 2024, Online Edition. Accessed 13 October 2026.

Dataset page: https://population.un.org/wpp/
Direct primary download: https://population.un.org/wpp/assets/Excel%20Files/1_Indicator%20(Standard)/CSV_FILES/WPP2024_Demographic_Indicators_Medium.csv.gz
Source SHA-256: 286ac36bb1415e2e1ade03acfef0a29f0e4c087e2f78e38c48f50c5df89082bc
Source size: 16,557,272 bytes; 84,360 rows including overlapping aggregates and projection years.

Data license: Creative Commons Attribution 3.0 IGO (CC BY 3.0 IGO): https://creativecommons.org/licenses/by/3.0/igo/
Primary UN licensing statement: https://population.un.org/dataportalapi/index.html
Primary WPP methodology and copyright: https://population.un.org/wpp/assets/Files/WPP2024_Methodology.pdf

Reuse and redistribution are allowed with attribution under the linked license. Attribute the UN source, link the license, identify modifications, and do not imply UN endorsement. No UN logo is included. Transformations here: selected years/locations/columns, changed units, renamed columns, added status, and charted the annual series. Chart and renderer prepared with Python, pandas and Pillow.

## Data dictionary and verification
- year: calendar year (births) and year containing the 1 July population estimate.
- location / un_location_id: UN location name and stable numerical identifier; world is 900.
- iso3: ISO alpha-3 code in the country companion file.
- births: estimated live births during the year, in people.
- population_july_1: estimated population as of 1 July, in people.
- status: `estimate` for all included rows.

Original source fields `Births` and `TPopulation1July` are in thousands. Values were multiplied by 1,000 and rounded to the nearest integer to remove binary floating-point noise; retain the source's modeled-estimate interpretation. No annual values were interpolated, smoothed, dropped or invented. The two chart panels use the same year spacing and different explicitly labeled units. The source CSV labels even historical rows `Variant=Medium`; classification uses the WPP historical cutoff of 2023, not this field. All projection years and the terminal 2101 row were excluded. Do not sum all 555 source locations: they include overlapping aggregates.


## Release files and checks
The world and country CSV bytes are unchanged from the reviewed package. Each
world year occurs once; every country/area has the full 1950–2023 series. All
rows are labelled estimate. World and country tables are separate: do not add
the World series to country totals. Country totals differ from the World row
by at most 21 births and 12 people after source-unit conversion and rounding.
No source error corrections, interpolation or additional imputation were made
for this release. The supplied approved chart is included unchanged.

The renderer expects DejaVu Sans regular and bold fonts in the Linux font path
shown in its source. It fetches and verifies the pinned UN source if absent.
The original full UN source is not bundled; validation.json records its identity.
RELEASE-MANIFEST.json records the hashes of the actual release files.

Download page: https://meplar.com/datasets/births-and-population/