Orchestrates a suite of validation checks on routine HF data. It standardizes column resolution, selects indicators, runs missing/duplicate/future/logic/ outlier checks, compiles a summary, and optionally translates and saves.

Usage

validate_routine_hf_data(
  data,
  id_col = "record_id",
  facility_col = "hf_uid",
  date_col = "date",
  yearmon_col = "yearmon",
  year_col = "year",
  month_col = "month",
  admin_cols = c("adm0", "adm1", "adm2", "adm3"),
  admin_guid_cols = c("adm0_guid", "adm1_guid", "adm2_guid", "adm3_guid"),
  core_id_cols = NULL,
  indicators = NULL,
  missing_vars = NULL,
  consistency_pairs = NULL,
  outlier_pairs = NULL,
  check_future_dates = TRUE,
  check_duplicates = TRUE,
  check_outliers = TRUE,
  check_facility_activeness = TRUE,
  hf_name_col = "hf",
  key_indicators = NULL,
  nonreport_window = 6,
  reporting_rule = "any_non_na",
  min_reporting_rate = 0.5,
  outlier_methods = c("iqr", "median", "mean", "consensus"),
  time_mode = "across_time",
  outlier_strictness = "balanced",
  sd_multiplier = 3,
  mad_constant = 1.4826,
  mad_multiplier = 9,
  iqr_multiplier = 2,
  min_n = 8,
  consensus_rule = 1,
  n_neighbour_impute = 5,
  output_path = NULL,
  output_name = "validation_of_hf_routine_data",
  output_formats = c("xlsx", "rds"),
  build_dictionary = FALSE,
  verbose = TRUE,
  language = "en"
)

Arguments

data: data.frame. Routine facility dataset.
id_col: character. Unique record id. Default "record_id".
facility_col: character. Facility id. Default "hf_uid".
date_col: character. Date column. Default "date".
yearmon_col: character. Year-month col. Default "yearmon".
year_col: character. Year col. Default "year".
month_col: character. Month col. Default "month".
admin_cols: character. Admin columns (adm0..adm3). Default c("adm0","adm1","adm2","adm3").
admin_guid_cols: character. Admin GUID columns. Default c("adm0_guid","adm1_guid","adm2_guid","adm3_guid").
core_id_cols: character|NULL. If NULL, uses the set above.
indicators: character|NULL. Numeric indicators to validate. If NULL, auto-detect from numeric columns after excluding core ids.
missing_vars: character|NULL. Additional variables to check for missing data beyond core_id_cols. Core ID columns are always checked. If NULL, checks both core_id_cols and indicators (default behavior). Use this to limit missing data checks to specific indicators while always including core IDs.
consistency_pairs: list|NULL. Each element is list(input=, output=). If NULL, defaults generated from common malaria cascade rules.
outlier_pairs: list|NULL. Pairs for outlier detection and correction. Structure: list(input = c("test"), output = c("conf")). Uses same format as consistency_pairs. The cascade rule is: input >= output. Both directions are checked automatically: (1) output outliers validated as output <= input, (2) input outliers validated as input >= output. Corrections use neighbor median. If NULL, defaults from malaria cascade.
check_future_dates: logical. Default TRUE.
check_duplicates: logical. Default TRUE.
check_outliers: logical. Default TRUE.
check_facility_activeness: logical. Check facility activeness. Default TRUE.
hf_name_col: character. Facility name column. Default "hf".
key_indicators: character|NULL. Key indicators for activeness check. If NULL, uses all indicators. Default NULL.
nonreport_window: integer. Minimum number of consecutive non-reporting months to classify a facility as inactive in method 3. Defaults to 6.
reporting_rule: character. Defines what counts as reporting for activeness: "any_non_na" (default, counts NA as non-reporting, 0 counts as reported) or "positive_only" (requires >0 value to count as reported).
min_reporting_rate: numeric. Minimum reporting rate threshold. Default 0.5.
outlier_methods: character. Any of c("iqr","median","mean").
time_mode: character. Time mode for outlier detection: "across_time", "within_year", or "seasonal". Seasonal compares same month across years. Default "across_time".
outlier_strictness: character. Outlier detection strictness: "balanced", "lenient", "strict", "advanced". Default "balanced".
sd_multiplier: numeric. Standard deviation multiplier for outlier detection. Default 3.
mad_constant: numeric. MAD constant for outlier detection. Default 1.4826.
mad_multiplier: numeric. MAD multiplier for outlier detection. Default 9.
iqr_multiplier: numeric. IQR multiplier for outlier detection. Default 2.
min_n: numeric. Minimum sample size for outlier detection. Default 8.
consensus_rule: numeric. Number of methods that must agree for consensus outlier flag. Default 1.
n_neighbour_impute: integer. Number of neighboring time periods (before and after) to use for computing the imputation median. Default 5 (uses 5 before + 5 after = 10 values).
output_path: character|NULL. If provided, results are saved to this path.
output_name: character. Base output name. Default "validation_routine_data_results".
output_formats: character. Any of c("xlsx","rds"). Default c("xlsx","rds").
build_dictionary: logical. Build data dictionary. Default TRUE.
verbose: logical. CLI messages. Default TRUE.
language: character. ISO-639-1 language code for output labels ("en", "fr", "pt"). Default "en".

Value

invisible named list with elements: Summary, Missing values, Missing values detail, Duplicate records, Future dates, Consistency summary, Consistency details, Outlier summary, Outlier detailed, HF activeness summary, HF activeness episodes, Data dictionary

Examples

if (FALSE) { # \dontrun{
validate_routine_hf_data(
  data = dhis2_data,
  indicators = c("conf","test","susp","maltreat")
)
} # }