Skip to contents

Process raw LQAS lots into classifications and district pass rates

Source: R/clean_lqas_im.R
process_lqas.Rd

Cleans the raw POLIS Lqas table with the same recipe as clean_afp() / clean_sia() – standardise names via the crosswalk, clean strings, parse and sanitise dates, fix admin names and (when a shape is supplied) reconcile admin GUIDs, then dedup by id – and rolls the cleaned lots up to a per-district-year table, classifying each lot two ways:

  • POLIS – the classification POLIS already ships in the download (lqas2_classification_name / lqas3_classification_name), derived from its unexposed REF_LQASThresholds lookup. This is the faithful answer.

  • derived – a transparent re-derivation from coverage (1 - children_found_unvaccinated / children_checked) against pass_threshold / warn_threshold, plus the documented default-60 and 2019 "sample size must be a multiple of 60 or INVALID" rules. Provided for QA comparison against the POLIS classes.

Usage

process_lqas(
  lqas,
  cfg = polis_active_config(),
  shape = NULL,
  default_checked = 60,
  multiple_of = 60,
  enforce_since = 2019,
  pass_threshold = 0.9,
  warn_threshold = 0.8,
  verbose = TRUE,
  summary = TRUE
)

Arguments

lqas

Raw LQAS table (data.frame/tibble), one row per lot.

cfg

A polis_config() object (default polis_config()); its crosswalk maps the raw column names to canonical snake_case.

shape

Optional district shape (an sf polygon layer or a long ADM2 attribute table) for admin GUID reconciliation via reconcile_admin_guids() (keyed on year). Default NULL.

default_checked

Lot size assumed when the sample size is missing (default 60, per POLIS).

multiple_of

Sample-size modulus enforced from enforce_since (default 60).

enforce_since

Year from which the multiple-of rule makes a lot INVALID in the derived classification (default 2019).

pass_threshold

Coverage (vaccinated fraction) at/above which a lot is a Pass in the derived classification (default 0.90).

warn_threshold

Coverage band for the derived 3-level "Intermediate" class (default 0.80). Lots in [warn_threshold, pass_threshold) are Intermediate (3-level) / Fail (2-level). Set to NULL to disable the Intermediate band, collapsing the derived classes to strict Pass/Fail.

verbose

Emit cli progress + the one-line roll-up result (default TRUE).

summary

Emit the full cli summary panel (rules + class breakdown). Default TRUE; run_pipeline() passes FALSE to keep the pipeline terse.

Value

A list with lots (the cleaned lot-level grain carrying both lqas2_polis / lqas3_polis and the derived coverage / invalid / lqas2_derived / lqas3_derived), district (per-district roll-up with n_lots, the POLIS n_pass_polis / n_fail_polis / n_invalid_polis / pass_pct_polis and the derived n_pass_derived / n_fail_derived / n_invalid_derived / pass_pct_derived), and meta.

Examples

lqas <- data.frame(
  Id = 1L, Admin0Name = "NIGERIA", Admin1Name = "KANO",
  Admin2Name = "NASSARAWA", Admin2GUID = "{A2}", Year = 2024L,
  ChildrenChecked = 60L, ChildrenFoundUnvaccinated = 3L,
  Lqas2ClassificationName = "Pass", Lqas3ClassificationName = "Pass"
)
process_lqas(lqas, verbose = FALSE)$district
#> # A tibble: 1 × 14
#>    year adm2_guid adm0    adm1  adm2      n_lots n_pass_polis n_fail_polis
#>   <int> <chr>     <chr>   <chr> <chr>      <int>        <int>        <int>
#> 1  2024 {A2}      NIGERIA KANO  NASSARAWA      1            1            0
#> # ℹ 6 more variables: n_invalid_polis <int>, n_pass_derived <int>,
#> #   n_fail_derived <int>, n_invalid_derived <int>, pass_pct_polis <dbl>,
#> #   pass_pct_derived <dbl>