This repository contains the implementation of GenSynthPop for the South-West district of The Hague, The Netherlands.

The synthetic population is generated for 14 neighborhoods and contains 10 individual-level attributes and 5 household-level attributes. The attributes are shown in the first column in the two tables in the Evaluation section below.

The synthetic population was generated as a case study for GenSynthPop, which was reported on in the paper GenSynthPop: Generating a Spatially Explicit Synthetic Population of Agents and Households from Aggregated Data¹. For a full report on the evaluation, please refer to that paper.

Neighborhoods

The fourteen neighborhoods are located in the South-West District of The Hague ('s Gravenhage), The Netherlands:

Neighborhood Code (CBS)	Neighborhood Name
BU05181785	Kerketuinen en Zichtenburg
BU05183284	Leyenburg
BU05183387	Venen, Oorden en Raden
BU05183396	Zijden, Steden en Zichten
BU05183398	Dreven en Gaarden
BU05183399	De Uithof
BU05183480	Morgenstond-Zuid
BU05183488	Morgenstond-West
BU05183489	Morgenstond-Oost
BU05183536	Zuiderpark
BU05183620	Moerwijk-Oost
BU05183637	Moerwijk-West
BU05183638	Moerwijk-Noord
BU05183639	Moerwijk-Zuid

The following map shows where in The Netherlands this synthetic population is located, and what the region looks like:

Data sets:

All data sets have been obtained through the Open Data portal of the Statistics Netherlands (Centraal Bureau voor de Statistiek or CBS in Dutch).

All data sources have been made available by CBS under the CC BY 4.0 license, as stated here.

1. marginal
   1. marginal_distributions_84583NED.csv. Source: Kerncijfers wijken en buurten 2019 (CBS StatLine):
   2. Regionale_kerncijfers_Nederland_19052024_185018.csv. Source: Regionale kerncijfers Nederland ( CBS StatLine)
2. individual
   1. gender
      1. gender_age-03759NED-formatted.csv. Source: Bevolking op 1 januari en gemiddeld; geslacht, leeftijd en regio (CBS StatLine)
   2. integer_age
      1. Leeftijdsopbouw Nederland 2019.csv. Source: Bevolkingspiramide (CBS)
   3. migration_background
      1. Bev__migratieachtergr__regio__2010_2022_29122023_115517.csv. Source: Bevolking; migratieachtergrond, generatie, lft, regio, 1 jan; 2010-2022 (CBS StatLine)
   4. education
      1. Primair_onderwijs__schoolregio_18012024_150722.csv. Source: (Speciaal) basisonderwijs en speciale scholen; leerlingen, schoolregio (CBS StatLine)
      2. Studenten__woonregio_2000_2022_18012024_150722.csv. Source: Leerlingen en studenten; onderwijssoort, woonregio 2000/'01-2022/' 23 (CBS StatLine)
      3. Bevolking__onderwijsniveau_en_herkomst_26012024_115001.csv. Source: Bevolking; hoogst behaald onderwijsniveau en herkomst (CBS StatLine)
   5. drivers_license
      1. Personen_met_rijbewijs__categorie__regio_19052024_184228.csv. Source: Personen met een rijbewijs; rijbewijscategorie, leeftijd, regio, 1 januari (CBS StatLine)
   6. household_position
      1. Huishoudens__personen__regio_26122023_151215.csv. Source: Huishoudens; personen naar geslacht, leeftijd en regio, 1 januari (CBS StatLine)
      2. Huishoudens__samenstelling__regio_25052024_174751.csv. Source: Huishoudens; samenstelling, grootte, regio, 1 januari (CBS StatLine)
3. household
   1. household_composition
      1. table_7ab235bf-b5a7-4077-bf56-3f5c8efec7d0.csv. Source: Groom usually older than bride (CBS)
      2. Marriages__key_figures_25052024_182843.csv. Source: Marriages and partnership registrations; key figures (CBS StatLine)
      3. Geboorte__kerncijfers_per_regio_25052024_182014.csv. Source: Geboorte; kerncijfers vruchtbaarheid, leeftijd moeder, regio (CBS StatLine)
   2. postal_code
      1. pc6hnr20190801_gwb.csv. Source: Buurt, wijk en gemeente 2019 voor postcode huisnummer (CBS):
   3. household_income
      1. Inkomen_huishoudens__kenmerken__regio_25052024_182249.csv. Source: Inkomen van huishoudens; huishoudenskenmerken, regio (indeling 2021) (CBS StatLine)
   4. vehicle_ownership
      1. Huishoudens_met_auto_of_motor__2010_2015_14062024_171657.csv. Source: Huishoudens in bezit van auto of motor; huishoudkenmerken, 2010-2015 (CBS StatLine):

Evaluation

Personal Level Attributes

			Z2		X2		absolute error
		DoF	Score	p-value	Score	p-value	total	standardized
age group	neighborhood	70	0.160028	1.000000	0.016845	1.000000	20.000000	0.000236
gender	neighborhood	28	0.168613	1.000000	0.185536	1.000000	26.000000	0.000306
	age group	10	0.612648	0.999983	0.579719	0.999987	150.559563	0.001773
integer age		106	2.144096	1.000000	3.601348	1.000000	124.606355	0.001468
	gender	212	26.086832	1.000000	31.153039	1.000000	1183.331065	0.013938
migration background	neighborhood	42	0.191206	1.000000	0.500781	1.000000	62.000000	0.000730
	age group	60	26.784031	0.999936	27.521184	0.999898	1175.066312	0.013844
	gender	6	1.421525	0.964536	1.140259	0.979733	296.952058	0.003498
	age group × gender	120	27.329810	1.000000	30.074286	1.000000	1195.071783	0.014080
absolved education		9	0.522445	0.999963	0.490611	0.999972	151.258870	0.001782
	neighborhood	42	0.066162	1.000000	0.237732	1.000000	43.868289	0.000517
	age group	171	69.205262	1.000000	71.689507	1.000000	1392.723678	0.016408
	gender	18	3.668375	0.999874	3.583914	0.999894	460.845914	0.005429
	age group × gender	342	73.542488	1.000000	78.444162	1.000000	1438.030022	0.016942
current education		18	5.271613	0.998370	5.559259	0.997681	241.257163	0.002842
	age group	738	400.828726	1.000000	88.304314	1.000000	344.817221	0.004062
	gender	36	5.406962	1.000000	6.937750	1.000000	265.580831	0.003129
	migration background	54	13.529318	1.000000	9.604675	1.000000	255.267884	0.003007
	absolved education	162	6.639501	1.000000	6.341646	1.000000	241.481890	0.002845
	age group × gender	1476	707.310600	1.000000	122.133588	1.000000	415.442018	0.004894
	age group × migration background	2214	853.547746	1.000000	108.171522	1.000000	392.444181	0.004624
	age group × absolved education	6642	464.316168	1.000000	92.685205	1.000000	360.126489	0.004243
	gender × migration background	108	20.653331	1.000000	13.456209	1.000000	281.248106	0.003313
	gender × absolved education	324	7.692929	1.000000	8.568989	1.000000	271.231784	0.003195
	migration background × absolved education	486	19.424479	1.000000	12.676304	1.000000	258.486825	0.003045
	age group × gender × migration background	4428	1574.162354	1.000000	157.970713	1.000000	487.830608	0.005747
	age group × gender × absolved education	13284	933.428873	1.000000	130.189879	1.000000	443.881705	0.005230
	age group × migration background × absolved education	19926	1093.462434	1.000000	113.877827	1.000000	416.662545	0.004909
	gender × migration background × absolved education	972	29.108628	1.000000	21.881322	1.000000	305.042419	0.003594
	age group × gender × migration background × absolved education	39852	2194.267693	1.000000	167.084847	1.000000	527.015116	0.006209
car license		2	0.000165	0.999917	0.000159	0.999921	3.595921	0.000042
	age	26	0.005861	1.000000	0.006811	1.000000	14.370524	0.000169
motor cycle license		2	0.002777	0.998612	0.001954	0.999024	6.370590	0.000075
	age	26	0.875771	1.000000	1.484794	1.000000	15.663175	0.000185
moped license		2	0.000220	0.999890	0.000196	0.999902	3.975836	0.000047
	age	26	0.003083	1.000000	0.010500	1.000000	13.535124	0.000159
	car license	4	0.002277	0.999999	0.003920	0.999998	3.975836	0.000047
	age × car license	52	0.027725	1.000000	0.075585	1.000000	13.535124	0.000159
household position		21	1.803299	1.000000	1.847752	1.000000	250.398797	0.002950
	neighborhood	42	1.802755	1.000000	2.501504	1.000000	146.000000	0.001720
	age group	420	249.202608	1.000000	126.578850	1.000000	754.259661	0.008886
	gender	42	2.699896	1.000000	2.918825	1.000000	336.681633	0.003967
	age group × gender	840	334.941882	1.000000	182.929638	1.000000	896.190081	0.010558

Household Attributes

			Z2		X2		absolute error
		DoF	Score	p-value	Score	p-value	total	standardized
household type	neighborhood	42	906.521255	0.000000	894.854325	0.000000	4098.000000	0.096605
postal code	neighborhood	2069	392.432681	1.000000	570.688983	1.000000	3199.000000	0.077825
income_group	age group	40	4.811664	1.000000	5.578801	1.000000	162.006477	0.003825
	income household type	40	0.858700	1.000000	1.054674	1.000000	138.199964	0.003263
	main bread winner migration background	30	1.270564	1.000000	1.394234	1.000000	164.570737	0.003886
	age group × income household type	160	21.969606	1.000000	32.654659	1.000000	269.499612	0.006364
	age group × main bread winner migration background	120	34.366320	1.000000	40.854727	1.000000	256.078370	0.006047
	income household type × main bread winner migration background	120	4.172994	1.000000	5.809870	1.000000	239.738984	0.005661
	age group × main bread winner migration background × income household type	480	57.803160	1.000000	76.238676	1.000000	420.321388	0.009925
cars	income household type	16	11.333855	0.788424	11.687847	0.765174	130.374345	0.003078
	hh size	20	2.453303	0.999999	2.831404	0.999998	138.374003	0.003267
	vehicle ownership income group	20	3.199855	0.999993	4.353822	0.999907	115.344156	0.002724
	car license	20	17.717134	0.606036	20.764026	0.411137	132.576966	0.003131
	income household type × hh size	80	10.263020	1.000000	11.653642	1.000000	148.560213	0.003508
	income household type × vehicle ownership income group	80	9.298031	1.000000	12.164716	1.000000	164.088294	0.003875
	income household type × car license	80	27.702845	1.000000	31.583354	1.000000	134.961108	0.003187
	hh size × vehicle ownership income group	100	6.117136	1.000000	8.300281	1.000000	170.569850	0.004028
	hh size × car license	100	18.241554	1.000000	25.092836	1.000000	154.187093	0.003641
	vehicle ownership income group × car license	100	26.546443	1.000000	34.161878	1.000000	156.713236	0.003700
	income household type × hh size × vehicle ownership income group	400	23.725567	1.000000	29.353700	1.000000	203.311026	0.004801
	income household type × hh size × car license	400	29.404266	1.000000	40.145015	1.000000	166.019409	0.003920
	income household type × vehicle ownership income group × car license	400	34.714700	1.000000	45.360212	1.000000	203.841026	0.004813
	hh size × vehicle ownership income group × car license	500	30.191990	1.000000	44.540901	1.000000	222.846215	0.005262
	income household type × hh size × vehicle ownership income group × car license	2000	54.483863	1.000000	73.215653	1.000000	265.690539	0.006274
motorcycles	income household type	16	3.331365	0.999661	3.876334	0.999100	37.931565	0.000896
	hh size	20	10.280284	0.962800	10.596614	0.956021	37.931565	0.000896
	vehicle ownership income group	20	2.359904	1.000000	2.954304	0.999996	35.458634	0.000837
	motorcycle license	12	1.270946	0.999947	1.348853	0.999926	35.458634	0.000837
	income household type × hh size	80	13.105943	1.000000	14.205072	1.000000	37.931565	0.000896
	income household type × vehicle ownership income group	80	7.924885	1.000000	11.107216	1.000000	44.933394	0.001061
	income household type × motorcycle license	48	3.537022	1.000000	4.063794	1.000000	37.931565	0.000896
	hh size × vehicle ownership income group	100	9.686371	1.000000	12.806302	1.000000	49.673925	0.001173
	hh size × motorcycle license	60	10.822410	1.000000	11.299931	1.000000	37.931565	0.000896
	vehicle ownership income group × motorcycle license	60	2.479304	1.000000	3.112846	1.000000	35.458634	0.000837
	income household type × hh size × vehicle ownership income group	400	16.255885	1.000000	19.531411	1.000000	50.254357	0.001187
	income household type × hh size × motorcycle license	240	13.667388	1.000000	14.965049	1.000000	37.931565	0.000896
	income household type × vehicle ownership income group × motorcycle license	240	8.115136	1.000000	11.201584	1.000000	44.933394	0.001061
	hh size × vehicle ownership income group × motorcycle license	300	10.301489	1.000000	13.574983	1.000000	49.673925	0.001173
	income household type × hh size × vehicle ownership income group × motorcycle license	1200	16.946660	1.000000	20.319974	1.000000	50.254357	0.001187

Using this repository

This repository uses Git LFS. Before cloning this repository, make sure to run

git lfs install

After cloning, create a virtual environment, and install the requirements:

python -m venv env
source env/bin/activate
pip install -r requirements.txt

First, the individual level attributes should be created:

python3 generate_individuals.py

This adds the configured attributes one at a time. After each attribute, the intermediate stage of the synthetic population is stored as both a CSV file and a Pandas DataFrame pickle. If these files already exist, the step is not executed again when the above code is run again. To make sure changes take place, remove files from output from the version and onwards where changes need to be applied.

The second step is the household generation:

python3 generate_households.py

This also runs in multiple steps, and again stores the intermediate results. In this step, both the synthetic population and the synthetic household may change at every stage, so both are stored for each stage, even if no changes occur.

Footnotes

1. Marco Pellegrino, Jan de Mooij, Tabea Sonnenschein et al. GenSynthPop: Generating a Spatially Explicit Synthetic Population of Agents and Households from Aggregated Data, 09 October 2023, PREPRINT (Version 1) available at Research Square [https://doi.org/10.21203/rs.3.rs-3405645/v1] ↩