Explainer
A synthetic population is read household by household
An invented neighbourhood of 100 people in 30 households shows how a synthetic population is read and how it is made. No number on this page is about Portugal.
The invented neighbourhood
- 0–14
- 15–24
- 25–64
- 65+
An invented neighbourhood: none of these people exists.
Each home is a household
30 households and 100 people, all invented. Choose a home to see who lives there.
View the homes as a table: The 30 invented households and the ages of who lives there
| Household | People | Ages |
|---|---|---|
| 1 | 1 | 48 |
| 2 | 2 | 68, 71 |
| 3 | 3 | 41, 39, 9 |
| 4 | 4 | 45, 43, 16, 12 |
| 5 | 5 | 44, 41, 8, 5, 70 |
| 6 | 5 | 52, 50, 22, 19, 81 |
| 7 | 1 | 79 |
| 8 | 2 | 29, 31 |
| 9 | 3 | 57, 55, 24 |
| 10 | 4 | 36, 34, 6, 2 |
| 11 | 5 | 47, 45, 17, 14, 76 |
| 12 | 5 | 42, 40, 11, 7, 3 |
| 13 | 1 | 69 |
| 14 | 2 | 74, 77 |
| 15 | 3 | 30, 28, 1 |
| 16 | 4 | 58, 56, 27, 23 |
| 17 | 5 | 43, 40, 13, 10, 68 |
| 18 | 5 | 67, 65, 37, 33, 4 |
| 19 | 1 | 86 |
| 20 | 2 | 46, 15 |
| 21 | 3 | 66, 65, 39 |
| 22 | 4 | 42, 44, 18, 15 |
| 23 | 5 | 45, 43, 12, 9, 6 |
| 24 | 5 | 55, 53, 29, 26, 72 |
| 25 | 1 | 25 |
| 26 | 2 | 73, 70 |
| 27 | 3 | 35, 33, 4 |
| 28 | 4 | 71, 49, 47, 20 |
| 29 | 5 | 49, 46, 21, 18, 80 |
| 30 | 5 | 31, 30, 5, 3, 0 |
View the ages as a table: The 100 invented people, by age
| Age | People |
|---|---|
| Ages 0–14 | 21 |
| Ages 15–24 | 12 |
| Ages 25–64 | 47 |
| Ages 65+ | 20 |
View the sizes as a table: The 100 invented people, by household size
| People in the household | Households | People |
|---|---|---|
| 1 | 5 | 5 |
| 2 | 5 | 10 |
| 3 | 5 | 15 |
| 4 | 5 | 20 |
| 5 | 10 | 50 |
After the fit, the neighbourhood is 2 people away from the table it had to match
In the published population, households are not written by hand like this neighbourhood’s: the model chooses them until the counts come as close as possible to the tables INE publishes for each parish. In miniature, with an invented table standing in for INE’s:
| Age | The table to match | 30 households at random | The neighbourhood, fitted |
|---|---|---|---|
| Ages 0–14 | 20 | 28 | 21 |
| Ages 15–24 | 12 | 9 | 12 |
| Ages 25–64 | 48 | 41 | 47 |
| Ages 65+ | 20 | 15 | 20 |
| Total | 100 | 93 | 100 |
| Away from the table | — | 23 | 2 |
“Away from the table” adds up the gaps group by group. Table and households invented for this example; in a real parish, the tables are INE’s (2021 Census).
Thirty households drawn at random were 23 people away from the table; the neighbourhood’s thirty, chosen for it, are 2 away. In the published population the fit is done parish by parish, against several of INE’s tables at once (age, marital status, education, work, among others), and it runs once: every published figure is one model run, with no margin. Each parish’s quality tier (A, B or C) says how close it came to its tables.
The published population is built in five steps
Learn the combinations. A model learns, from the 2021 Census public-use sample, which people and households tend to occur together: ages, education, work, who lives with whom.
Generate candidates and fit them to INE’s tables. For each parish the model generates thousands of possible households and gives more weight to those that make the counts match the tables INE published for it: how many people of each age, each education level, in homes of each size. It is the fit in the table above, with many tables at once.
Draw whole households. Whole households are drawn from the candidates, like the homes in the neighbourhood: nobody is split.
Add the residents of care homes. People in care homes and other collective quarters are added from INE’s counts.
Publish with a quality tier. Every published column has a declared meaning, and each parish comes out with a tier (A, B or C), which combines how close it came to the tables and the number of residents.
A real parish reads the same way
one model run, no marginGenerated people and households, not real people, families, or addresses.
Every 2021 Census parish has its own generated population, fitted to INE’s tables for that parish, with the same questions: who lives alone, ages, work, families.