Skip to contents

Convert a Synthetic Population to a Contingency Table

Source: R/fn_utils-extractors.R
synthetic_population_to_contingency.Rd

Converts an agent-level synthetic population into a contingency table.

Usage

synthetic_population_to_contingency(
  df_synthetic_population,
  columns = NULL,
  full_crosstab = FALSE
)

Arguments

df_synthetic_population

A synthetic population stored as a data.frame or data.table.

columns

Character vector identifying the variables to include in the contingency table.

If NULL, all available variables are used.

full_crosstab

Logical value indicating whether all possible combinations of factor levels should be represented.

If:

FALSE

Only observed combinations are returned.

TRUE

Missing combinations are included with count = 0.

Value

A contingency table containing the supplied grouping variables and a count column.

Details

The resulting table contains one row for each unique combination of the supplied attributes together with a count column indicating the number of synthetic agents belonging to that group.

This function is one of the core analytical utilities in replica and is used for:

  • Validation of synthetic populations.

  • Comparison with reference contingency tables.

  • Goodness-of-fit assessment.

  • Calculation of marginal distributions.

  • Python-parity testing.

The function aggregates the synthetic population by the supplied variables and counts the number of agents in each resulting group.

When full_crosstab = TRUE, a complete cross-classification of all observed factor levels is generated and any absent combinations receive a count of zero.

This behaviour is particularly useful when comparing synthetic populations against reference distributions.

See also

validate_synthetic_population_fit, calculate_z_squared_score, prepare_contingency_table

Examples

population <- data.frame(
  gender = c(
    "Male",
    "Male",
    "Female"
  ),
  education = c(
    "Degree",
    "Degree",
    "School"
  )
)

synthetic_population_to_contingency(
  population,
  c(
    "gender",
    "education"
  )
)
#>   gender education count
#> 1   Male    Degree     2
#> 2 Female    School     1

synthetic_population_to_contingency(
  population,
  c(
    "gender",
    "education"
  ),
  full_crosstab = TRUE
)
#>   gender education count
#> 1 Female    Degree     0
#> 2 Female    School     1
#> 3   Male    Degree     2
#> 4   Male    School     0