Constructing synthetic populations in the age of big data

Table 1 Frequencies (in per cent) of categorical variables in the original confidential population and in the synthetic population

Variables	Original confidential population	Synthetic population
Variables	Frequencies	Frequencies
Lung cancer
No	99.86	99.88
Yes	0.14	0.12
Pancreas cancer
No	99.99	99.99
Yes	0.01	0.01
Main source of income
Employee	47	47
Civil servant	7.4	7.4
Salary as company director	2.4	2.4
Other income from labour	0.3	0.3
Income as company owner	14.7	14.7
Income from property	0.4	0.4
Unemployment benefits	1	1
Disability pension	2.9	2.9
Retirement pension	17.8	17.8
Social assistance benefits	3.2	3.2
Other social security	1	1
Study grant	0.8	0.8
Other	0.1	0.1
No income	1	1
Household size (number of persons)
1	17.5	17.2
2	29.7	30.2
3	16.4	16.7
4	23.1	22.4
5	9.4	9.4
6 and more	3.9	4.1
Migration background
Dutch	78.9	78.7
Moroccan	2.2	2.2
Turkish	2.4	2.4
Surinam	2.1	2.1
Netherlands Antilles and Aruba	0.9	0.9
Other non-Western	4.2	4.2
Other Western	9.4	9.5
Type of household
Institutional	1.4	1.6
Non-institutional	98.6	98.4

* Hospital admission data from the National Medical Registry (LRM), primary care data from the Netherlands Institute for Health Services Research (NIVEL), drug reimbursement data from the Dutch registry on medication use (Medicijntab), all provided through Statistics Netherlands

ISSN: 1478-7954