Explicitly Create an Experimental DSIR Data Object

Adds a dsir_data tibble subclass to an already cleaned indicator table. Cleaners return this subclass by default; output = "tibble" opts out. Downloads and directory functions retain their existing return types. This experimental interface and its metadata schema may change.

Usage

as_dsir_data(x)

# S3 method for class 'dsir_data'
print(x, ...)

# S3 method for class 'dsir_data'
summary(object, ...)

Arguments

x: A data frame with the typed 15-column schema returned by gho_clean(), ghe_clean() or sdg_clean(). Optional columns are kept.
...: Arguments passed to the underlying tibble printer, or reserved for future summary options.
object: A dsir_data object.

Value

A tibble inheriting from dsir_data, with a dsir_metadata attribute. Use dsir_metadata() to inspect it.

Details

Conversion does not retrieve data or recover discarded fields. Retain dimensions and metadata when cleaning if they are needed for analysis. Recorded who_provenance is moved into the metadata, not duplicated. Unknown retrieval times and definitions remain unknown. The initial as_supplied status describes the input; it does not certify its origin, completeness, units or statistical validity.

Row selection and ordering preserve the class. Column modifications, including dplyr::mutate() and base replacement, preserve it while the core schema remains valid. New columns are marked derived. Changes to input fields mark their metadata as needing review; changes to values, units or identifiers also flag related interpretation. This records field changes, not an executable transformation history or inferred formulas. Deleting a core column or changing its required type returns a plain tibble. Removing optional input columns records the loss of information.

Joins, row binding, persistent grouping and explicit conversion with tibble::as_tibble() or as.data.frame() return ordinary tables without object-level provenance. Per-row source and dimension columns remain. Use .by for transient dplyr grouping. Summary operations create new ordinary tables. Direct manipulation of attributes is outside this experimental contract.

data_snapshot() accepts an unchanged GHO object with the required detailed fields and original provenance. Row-selected objects need an explicit query scope. Objects whose metadata needs review are rejected; retain the original observations for reproducible source snapshots.

summary() returns a structured list with current coverage, indicator identities, dimension values, per-column missing counts and provenance. Numeric diagnostics count missing and non-finite value_num entries. This first version does not pool numerical statistics across indicators, countries, units or demographic groups. Empty list entries and entries containing only missing values count as missing in list columns.

Examples

raw <- tibble::tibble(
  IndicatorCode = "EXAMPLE", IndicatorName = "Example indicator",
  SpatialDim = c("PHL", "LAO"), TimeDim = c(2022L, 2023L),
  NumericValue = c(10, 20), Value = c("10", "20")
)
x <- as_dsir_data(gho_clean(raw))
x
summary(x)
x$value_num <- x$value_num / 100
dsir_metadata(x)$processing