From 9267f78735ff107c3a5886d85ef865691e26f806 Mon Sep 17 00:00:00 2001
From: your name <eli.pousson@gmail.com>
Date: Wed, 22 May 2024 10:49:51 -0400
Subject: [PATCH] Add na_zero parameter to collapse_acs_variables

---
 R/collapse_acs_variables.R    | 63 ++++++++++++++++++++++++-----------
 man/collapse_acs_variables.Rd | 10 ++++--
 2 files changed, 51 insertions(+), 22 deletions(-)

diff --git a/R/collapse_acs_variables.R b/R/collapse_acs_variables.R
index 31640f6..483fd8d 100644
--- a/R/collapse_acs_variables.R
+++ b/R/collapse_acs_variables.R
@@ -15,9 +15,13 @@
 #' @param value_col,moe_col Value and margin of error column names (default to
 #'   "estimate" and "moe").
 #' @param na.rm Passed to [sum()], Default: `TRUE`
+#' @param na_zero If `TRUE`, and the collapsed sum of a MOE is 0, replaced MOE
+#'   value with `NA`. This is beneficial for percent estimates with the margin
+#'   of error falls below 1% and is rounded to 0 with the default number of
+#'   digits.
 #' @param digits Passed to [round()], Default: 2
-#' @param extensive Must be `TRUE`. If `FALSE`, summarize collapsed variables
-#'   using a weighted mean. This is a planned feature.
+#' @param extensive Must be `TRUE`. If `FALSE` (not currently supported),
+#'   summarize collapsed variables using a weighted mean.
 #' @examples
 #' \dontrun{
 #' if (interactive()) {
@@ -55,6 +59,7 @@ collapse_acs_variables <- function(data,
                                    value_col = "estimate",
                                    moe_col = "moe",
                                    na.rm = TRUE,
+                                   na_zero = TRUE,
                                    digits = 2,
                                    .add = FALSE,
                                    extensive = TRUE) {
@@ -107,10 +112,30 @@ collapse_acs_variables <- function(data,
     value_col = value_col,
     moe_col = moe_col,
     na.rm = na.rm,
+    na_zero = na_zero,
     digits = digits
   )
 }
 
+moe_sum_safe <- function(moe,
+                         estimate,
+                         na.rm = TRUE,
+                         na_zero = TRUE,
+                         digits = 2) {
+  dplyr::if_else(
+    all(is.na(moe)) | (na_zero & sum(moe) == 0),
+    NA_real_,
+    round(
+      tidycensus::moe_sum(
+        moe,
+        estimate = estimate,
+        na.rm = na.rm
+      ),
+      digits = digits
+    )
+  )
+}
+
 #' Summarise ACS data assuming extensive variables
 #'
 #' @importFrom dplyr summarise across
@@ -124,6 +149,7 @@ summarise_acs_extensive <- function(
     value_col = "estimate",
     moe_col = "moe",
     na.rm = TRUE,
+    na_zero = TRUE,
     digits = 2) {
   perc_cols <- acs_perc_cols(
     value_col = value_col,
@@ -140,12 +166,11 @@ summarise_acs_extensive <- function(
         sum(.data[[value_col]], na.rm = na.rm),
         digits = digits
       ),
-      "{moe_col}" := round(
-        tidycensus::moe_sum(
-          .data[[moe_col]],
-          estimate = .data[[value_col]],
-          na.rm = na.rm
-        ),
+      "{moe_col}" := moe_sum_safe(
+        .data[[moe_col]],
+        .data[[value_col]],
+        na.rm = na.rm,
+        na_zero = na_zero,
         digits = digits
       ),
       dplyr::across(
@@ -169,24 +194,22 @@ summarise_acs_extensive <- function(
       sum(.data[[value_col]], na.rm = na.rm),
       digits = digits
     ),
-    "{moe_col}" := round(
-      tidycensus::moe_sum(
-        .data[[moe_col]],
-        estimate = .data[[value_col]],
-        na.rm = na.rm
-      ),
+    "{moe_col}" := moe_sum_safe(
+      .data[[moe_col]],
+      .data[[value_col]],
+      na.rm = na.rm,
+      na_zero = na_zero,
       digits = digits
     ),
     "{perc_cols[[1]]}" := round(
       sum(.data[[perc_cols[[1]]]], na.rm = na.rm),
       digits = digits
     ),
-    "{perc_cols[[2]]}" := round(
-      tidycensus::moe_sum(
-        .data[[perc_cols[[2]]]],
-        estimate = .data[[perc_cols[[1]]]],
-        na.rm = na.rm
-      ),
+    "{perc_cols[[2]]}" := moe_sum_safe(
+      .data[[perc_cols[[2]]]],
+      .data[[perc_cols[[1]]]],
+      na.rm = na.rm,
+      na_zero = na_zero,
       digits = digits
     ),
     dplyr::across(
diff --git a/man/collapse_acs_variables.Rd b/man/collapse_acs_variables.Rd
index 201cb15..255d31e 100644
--- a/man/collapse_acs_variables.Rd
+++ b/man/collapse_acs_variables.Rd
@@ -14,6 +14,7 @@ collapse_acs_variables(
   value_col = "estimate",
   moe_col = "moe",
   na.rm = TRUE,
+  na_zero = TRUE,
   digits = 2,
   .add = FALSE,
   extensive = TRUE
@@ -40,6 +41,11 @@ column added to the returned data frame.}
 
 \item{na.rm}{Passed to \code{\link[=sum]{sum()}}, Default: \code{TRUE}}
 
+\item{na_zero}{If \code{TRUE}, and the collapsed sum of a MOE is 0, replaced MOE
+value with \code{NA}. This is beneficial for percent estimates with the margin
+of error falls below 1\% and is rounded to 0 with the default number of
+digits.}
+
 \item{digits}{Passed to \code{\link[=round]{round()}}, Default: 2}
 
 \item{.add}{When \code{FALSE}, the default, \code{group_by()} will
@@ -50,8 +56,8 @@ This argument was previously called \code{add}, but that prevented
 creating a new grouping variable called \code{add}, and conflicts with
 our naming conventions.}
 
-\item{extensive}{Must be \code{TRUE}. If \code{FALSE}, summarize collapsed variables
-using a weighted mean. This is a planned feature.}
+\item{extensive}{Must be \code{TRUE}. If \code{FALSE} (not currently supported),
+summarize collapsed variables using a weighted mean.}
 }
 \description{
 \code{\link[=collapse_acs_variables]{collapse_acs_variables()}} uses \code{\link[forcats:fct_collapse]{forcats::fct_collapse()}} to aggregated