Download, cache, collate, filter and extrapolate UK Population estimates and projections
ukpopulation: UK Demographic Projections
The statistical agencies of the United Kingdom, that is: ONS, StatsWales, NR Scotland, and NISRA, all produce annual population estimates and projection data. Although the data are essentially the same, the quantity, format, and availability varies between agencies and datasets. All of the projection data is available by (single year of) age and gender.
National population projections (NPP) are the responsibility of ONS who provide the data for each country within the UK, including 15 variants covering a number of possible future scenarios. The current data is based on 2016 population estimates and project a century to 2116.
Subnational population projections (SNPP) are the responsibility of each country's agencies (ONS for England), and project 25 years from a base year that depends on the country in question:
|Country||Latest SNPP year range (as of June 2018)|
Mid-year population estimates (MYE) are available for the entire UK by local authority, single year of age and gender, from 1991 to 2016 inclusive.
The countries within the UK produce their own SNPP data, and also produce some (patchy) variant projections. The ONS currently regard these (the England ones at least) as "experimental".
|High life expectancy||php||x||x|
|Low life expectancy||plp||x||x|
|Moderately high life expectancy||pjp||x|
|Moderately low life expectancy||plp||x|
|0% future EU migration||ppq||x|
|50% future EU migration||ppr||x|
|150% future EU migration||pps||x|
|Zero net migration||ppz||x||x||x||x|
|Young age structure||hlh|
|Old age structure||lhl|
|No mortality improvement||pnp|
|Long term balanced net migration||ppb||x|
Nomisweb provides an API which allows relatively easy programmatic access the to data, and by far the preferred source of data. Currently not all the data is available from this source but this may change.
Nomisweb currently hosts the ONS principal NPP data for the UK, the SNPP data for England, and all of the MYE data.
All other data: ONS NPP variants, SNPP data for Wales, Scotland and Northern Ireland are available in different formats from the appropriate agency's website.
The purpose of this package is to provide a unified interface to both SNPP and NPP data, including variants:
- encapsulating the downloading, processing and caching of the NPP and SNPP data from the various sources.
- consistently differentiating by age (single year, up to 90) and gender over the various datasets.
- providing a unified format for all the data.
- providing a method of synthesising SNPP variant projections using SNPP principal and NPP principal/variant projections
- providing a method of extrapolating SNPP data using NPP data
- enabling easy filtering and aggregating of of the data, e.g. extracting projections of the working-age population.
Methodology and Detail
- Nomisweb: UK NPP by country/age/gender, England SNPP by LAD/age/gender, UK MYE by LAD/age/gender.
- ONS: UK NPP variants by country/age/gender.
- Stats Wales: Wales SNPP by LAD/age/gender.
- National Records of Scotland: Scotland SNPP by LAD equivalent/age/gender.
- Northern Ireland Statistics and Research Agency: Northern Ireland SNPP by LAD equivalent/age/gender.
- Note that the SNPP for each country, and the NPP data may not have the same reference year. (See table above).
- NPP data is broken down by country (England/Wales/Scotland/Northern Ireland), for all the variant projections indicated in the second table above.
- Column headings and category values follow the nomisweb/census conventions:
GEOGRAPHY_CODE: ONS country, LAD, or LAD-equivalent code
GENDER: 1=Male, 2=Female
C_AGE: 0-90, where 90 represents 90 or over. To avoid ambiguity, this is an exception - nomisweb census values are typically age+1)
OBS_VALUE: count of persons
- All data are cached for swift retrieval.
The SNPP data can be extrapolated using the longer-term NPP data. This is done independently for each age and gender in order to try to capture the age-gender structure of the original population. Aggregation only takes place on the extrapolated age-gender specific values. This means that the trends shown by SNPP geographies with different age-gender structures will differ.
If, conversely, the extrapolation was done on the aggregated populations for each SNPP geography, then each SNPP geography would have an identical trend, which would be identical to that of the NPP data.
This methodology can be more formally explained by the following equation for the aggregate SNPP S(g,y) for a given geography and year:
where N is the NPP, a is age, s is gender, y bar is a reference year (typically the final year in the SNPP data), and c(g) represents a mapping from a SNPP geography (LAD) to a NPP one (country).
Projection of Variants
The extrapolation methodology above can equally be applied to synthesising SNPP variants from SNPP principal and NPP variant data. The equivalent expression to the above is:
where the subscripts V and 0 refer to the variant and the principal projections respectively.
Requires Python 3.5 or higher. Dependencies should resolve automatically, but if not see troubleshooting
$ python3 -m pip install ukpopulation
$ conda install ukpopulation
This assumes you have added the conda-forge channel, which can be done with
$ conda config --add channels conda-forge
Some of the examples (see below) plot graphs and have a dependency on
matplotlib, which can be installed with either pip or conda as appropriate.
To clone the repo and install locally:
$ git clone https://github.com/nismod/ukpopulation $ ./setup.py install
(substituting the URL if a fork has been taken). The test data cache directory contains a file
NOMIS_API_KEY which defines a dummy key for testing purposes only. The test suite can be run from the (project root dir) using:
$ ./setup.py test
Ensure you are using the correct version (>=3) of pip:
$ pip --version pip 9.0.1 from /usr/lib/python3/dist-packages (python 3.6)
If not replace
python3 -m pip
If the installation has missing dependencies, try:
$ pip install -r requirements.txt $ ./setup.py install
If (with python 3.5?) you encounter
AttributeError: module 'html5lib.treebuilders' has no attribute '_base'
$ pip install html5lib=0.9999999
should fix it. But better solution is to upgrade to python3.6
If matplotlib fails to install due to a missing dependency (tkinter), this can be fixed on Debian variants by
$ sudo apt install python3-tk
If your problem isn't addressed above, please post an issue including as much supporting information as possible.
Retrieve SNPP for specific LADs
This example fetches the 2018 projection for Newcastle by gender and age.
>>> import ukpopulation.snppdata as SNPPData >>> snpp = SNPPData.SNPPData()
Cache directory: ./raw_data/ using cached LAD codes: ./raw_data/lad_codes.json Collating SNPP data for England... ./raw_data/NM_2006_1_metadata.json found, using cached metadata... Using cached data: ./raw_data/NM_2006_1_56aba41fc0fab32f58ead6ae91a867b4.tsv ./raw_data/NM_2006_1_metadata.json found, using cached metadata... Using cached data: ./raw_data/NM_2006_1_dbe6c087fb46306789f7d54b125482e4.tsv Collating SNPP data for Wales... Collating SNPP data for Scotland... Collating SNPP data for Northern Ireland...
>>> newcastle=snpp.filter("E08000021", 2018) >>> newcastle.head()
C_AGE GENDER GEOGRAPHY_CODE OBS_VALUE PROJECTED_YEAR_NAME 0 0 1 E08000021 1814.0 2018 1 1 1 E08000021 1780.0 2018 2 2 1 E08000021 1770.0 2018 3 3 1 E08000021 1757.0 2018 4 4 1 E08000021 1747.0 2018
This example fetches the total population projections for Newcastle from 2018 to 2039.
>>> import ukpopulation.snppdata as SNPPData >>> snpp = SNPPData.SNPPData()
Cache directory: ./raw_data/ using cached LAD codes: ./raw_data/lad_codes.json Collating SNPP data for England... ./raw_data/NM_2006_1_metadata.json found, using cached metadata... Using cached data: ./raw_data/NM_2006_1_56aba41fc0fab32f58ead6ae91a867b4.tsv ./raw_data/NM_2006_1_metadata.json found, using cached metadata... Using cached data: ./raw_data/NM_2006_1_dbe6c087fb46306789f7d54b125482e4.tsv Collating SNPP data for Wales... Collating SNPP data for Scotland... Collating SNPP data for Northern Ireland... ```python >>> newcastle=snpp.aggregate(["GENDER", "C_AGE"], "E08000021", range(2018,2039)) >>> newcastle.head()
GEOGRAPHY_CODE PROJECTED_YEAR_NAME OBS_VALUE 0 E08000021 2018 299132.0 1 E08000021 2019 300530.0 2 E08000021 2020 301699.0 3 E08000021 2021 302729.0 4 E08000021 2022 303896.0
Retrieve NPP data filtered by age
Here's how to get the total working-age population by country from 2016 to 2050:
>>> import ukpopulation.nppdata as NPPData >>> npp = NPPData.NPPData()
Cache directory: ./raw_data/ using cached LAD codes: ./raw_data/lad_codes.json Loading NPP principal (ppp) data for England, Wales, Scotland & Northern Ireland ./raw_data/NM_2009_1_metadata.json found, using cached metadata... Using cached data: ./raw_data/NM_2009_1_444caf1f672f0646722e389963289973.tsv
>>> uk_working_age=npp.aggregate(["GENDER", "C_AGE"], "ppp", NPPData.NPPData.UK, range(2016,2051), ages=range(16,75)) >>> uk_working_age.head() GEOGRAPHY_CODE PROJECTED_YEAR_NAME OBS_VALUE 0 E92000001 2016 40269470 1 E92000001 2017 40460118 2 E92000001 2018 40591965 3 E92000001 2019 40704521 4 E92000001 2020 40834471
And this aggregates the figures for Great Britain:
>>> gb_working_age=npp.aggregate(["GEOGRAPHY_CODE", "GENDER", "C_AGE"], "ppp", NPPData.NPPData.GB, range(2016,2051), ages=range(16,75)) >>> gb_working_age.head() PROJECTED_YEAR_NAME OBS_VALUE 0 2016 46590014 1 2017 46801693 2 2018 46944219 3 2019 47063069 4 2020 47201882
NB SNPP data can also be filtered by age and/or gender and/or geography in the same way.
Retrieve NPP variants for England & Wales
First detailed data (by age, gender and country), then aggregated by age and gender.
>>> import ukpopulation.nppdata as NPPData >>> npp=NPPData.NPPData() Cache directory: ./raw_data/ using cached LAD codes: ./raw_data/lad_codes.json Loading NPP principal (ppp) data for England, Wales, Scotland & Northern Ireland ./raw_data/NM_2009_1_metadata.json found, using cached metadata... Using cached data: ./raw_data/NM_2009_1_444caf1f672f0646722e389963289973.tsv >>> high_growth = npp.detail("hhh", NPPData.NPPData.EW) >>> high_growth.head() C_AGE GENDER OBS_VALUE PROJECTED_YEAR_NAME GEOGRAPHY_CODE 0 0 1 343198 2016 E92000001 1 0 1 334025 2017 E92000001 2 0 1 345332 2018 E92000001 3 0 1 349796 2019 E92000001 4 0 1 354274 2020 E92000001 >>> high_growth_agg = npp.aggregate(["GENDER", "C_AGE"], "hhh", NPPData.NPPData.EW) >>> high_growth_agg.head() GEOGRAPHY_CODE PROJECTED_YEAR_NAME OBS_VALUE 0 E92000001 2016 55268067 1 E92000001 2017 55660155 2 E92000001 2018 56115027 3 E92000001 2019 56568795 4 E92000001 2020 57019007 >>>
Extrapolate MYE using SNPP and NPP data
Construct aggregate data for Exeter from 2011-2065:
- use MYE data up to 2016, aggregated by age and gender.
- then use SNPP data up to 2041, aggregated by age and gender.
- extrapolate using NPP data and Exeter's (2041) age-gender structure.
- aggregrate the extrapolated data by age and gender
- plot the data.
In this example we extrapolate and aggregrate the SNPP for every LAD in Wales:
- for each area,
- extrapolate from 2039 to 2050 using the 2039 age-gender structure.
- aggregate the extrapolated datma by age and gender.
- append to full dataset.
- save Wales dataset as csv:
Construct an SNPP variant by applying NPP variant to a specific LAD
Here we apply the "hhh" (high growth) and "lll" (low growth) NPP variants to the SNPP data for Newcastle:
- calculate the principal ("ppp") projection by simply aggregrating the SNPP data for Newcastle, 2018-2039, by age and gender.
- calculate the variants by weighting the unaggregated data (i.e. by age and gender) by the ratio of the NPP variant/principal.
- aggregrate the variant data by age and gender.
- plot the results.
Extrapolating an SNPP variant
Here we build on the examples above by not only applying the NPP variant, but extrapolating too. The process first involves extrapolating the SNPP by the NPP principal variant. The extrapolated data then has the variant adjustments applied to it.
Package documentation can be viewed like so:
import ukpopulation.myedata as MYEData help(MYEData) import ukpopulation.nppdata as NPPData help(NPPData) import ukpopulation.snppdata as SNPPData help(SNPPData)
This package was developed as a component of the EPSRC-funded MISTRAL programme, part of the Infrastructure Transitions Research Consortium.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size ukpopulation-1.0.3-py3.6.egg (32.2 kB)||File type Egg||Python version 3.6||Upload date||Hashes View hashes|
|Filename, size ukpopulation-1.0.3-py3-none-any.whl (18.6 kB)||File type Wheel||Python version py3||Upload date||Hashes View hashes|
|Filename, size ukpopulation-1.0.3.tar.gz (26.3 kB)||File type Source||Python version None||Upload date||Hashes View hashes|
Hashes for ukpopulation-1.0.3-py3-none-any.whl