replace missing values with "" in receipts

process/R/out_receipts.R

@@ -15,6 +15,7 @@ out_receipts_payer <- function(
 
   receipts_payer <-
     receipts |>
+    receipts_replace_na() |>
     distinct(
       org_name,
       is_org,
@@ -35,6 +36,20 @@ out_receipts_payer <- function(
   dirname(out)
 }
 
+receipts_replace_na <- function(receipts) {
+  receipts |>
+    tidyr::replace_na(list(
+      org_name = "",
+      profession = "",
+      employers_name = "",
+      street_1 = "",
+      city = "",
+      state = "",
+      full_zip = "",
+      country_name = ""
+    ))
+}
+
 out_receipts <- function(
   path_data_prep_receipts,
   path_out_receipts_payer,
@@ -59,14 +74,16 @@ out_receipts <- function(
 
   receipts <-
     receipts |>
+    receipts_replace_na() |>
     left_join(receipts_payer, by = cols_payer_common) |>
-    select(-any_of(cols_payer_common)) |>
+    select(-any_of(cols_payer_common), -name_sort) |>
     mutate(
       is_donation = receipt_type_code %in% c("IND", "CPCM", "GEN", "PPTY", "OUTS", "NFPC"),
       .after = receipt_type_code
     ) |>
     relocate(payer_id, .after = report_id) |>
-    collect()
+    collect() |>
+    mutate(across(occur_date, lubridate::mdy))
 
   arrow::write_parquet(receipts, out)
 

process/R/prepare_candidates.R

@@ -113,7 +113,8 @@ fastlink_candidates <- function(candidates_for_matching, candidate_listing_for_m
     varnames = c("name_clean", "street", "city"),
     stringdist.match = c("name_clean", "street"),
     partial.match = c("name_clean", "street"),
-    stringdist.method = "dl"
+    stringdist.method = "dl",
+    threshold.match = 0.9
   )
 }
 
@@ -128,12 +129,12 @@ fastlink_match_candidates <- function(
     candidates_for_matching,
     candidate_listing_for_matching,
     candidates_linked,
-    threshold.match = 0.8
+    threshold.match = 0.9
   )
 
   matches <- as_tibble(matches)
 
-  select(matches, sboe_id, candidate_id)
+  distinct(matches, sboe_id, candidate_id)
 }
 
 candidates_match <- function(