Skip to contents

Execute estimation workflow for surveys

Source: R/workflow.R
workflow.Rd

This function executes a sequence of statistical estimations on Survey objects, applying functions from the R survey package with appropriate metadata. Automatically handles different survey types and periodicities.

Usage

workflow(svy, ..., estimation_type = "monthly")

Arguments

svy

A list of Survey objects, or a PoolSurvey. Even for a single survey, wrap it in list(): workflow(svy = list(my_survey), ...). Must contain properly configured sample design.

...

Calls to survey package functions (such as svymean, svytotal, svyratio, etc.) that will be executed sequentially

estimation_type

Type of estimation (default "monthly") that determines which weight to use. Options: "monthly", "quarterly", "annual", or vector with multiple types

Value

data.table with results from all estimations, including columns:

  • stat: Name of estimated statistic

  • value: Estimation value

  • se: Standard error

  • cv: Coefficient of variation

  • estimation_type: Type of estimation used

  • survey_edition: Survey edition

  • Other columns depending on estimation type

Details

The function automatically selects the appropriate sample design according to the specified estimation_type. For each Survey in the input list, it executes all functions specified in ... and combines the results.

Supported estimation types:

  • "monthly": Monthly estimations

  • "quarterly": Quarterly estimations

  • "annual": Annual estimations

For PoolSurvey objects, it uses a specialized methodology that handles pooling of multiple surveys.

See also

svymean for population means svytotal for population totals svyratio for ratios svyby for domain estimations PoolSurvey for survey pooling

Other workflows: RecipeWorkflow-class, evaluate_cv(), print.RecipeWorkflow(), publish_workflow(), read_workflow(), reproduce_workflow(), save_workflow(), workflow_from_list(), workflow_table()

Examples

# Simple estimation with a test survey
dt <- data.table::data.table(
  x = rnorm(100), g = sample(c("a", "b"), 100, TRUE),
  w = rep(1, 100)
)
svy <- Survey$new(
  data = dt, edition = "2023", type = "test",
  psu = NULL, engine = "data.table",
  weight = add_weight(annual = "w")
)
result <- workflow(
  svy = list(svy),
  survey::svymean(~x, na.rm = TRUE),
  estimation_type = "annual"
)

# \donttest{
# ECH example with domain estimations
# result <- workflow(
#   survey = list(ech_2023),
#   svyby(~unemployed, ~region, svymean, na.rm = TRUE),
#   estimation_type = "annual")
# }