From d953edce545792c5d3987ddff3ab59d511e15208 Mon Sep 17 00:00:00 2001
From: Rob Wiederstein <khuon68@gmail.com>
Date: Thu, 16 Apr 2026 06:25:34 -0400
Subject: [PATCH] Fix listing coordinates using building footprint geometry

RentCast geocoding is approximate and placed 878 Chalmers in the lake.
Override RentCast lat/lng with building centroid coords from owners.rds
when address matches, giving accurate per-structure placement.
---
 data-raw/update_listings.R |  46 +++++++++++++++++++++++++++++++++----
 data/listings.rds          | Bin 637 -> 693 bytes
 2 files changed, 42 insertions(+), 4 deletions(-)

diff --git a/data-raw/update_listings.R b/data-raw/update_listings.R
index a5b0ba2..3d37460 100644
--- a/data-raw/update_listings.R
+++ b/data-raw/update_listings.R
@@ -51,13 +51,51 @@ in_plat  <- lengths(sf::st_within(listings_sf, plats)) > 0
 listings <- listings_raw[in_plat, ]
 cat("After plat clip:", nrow(listings), "listings\n")
 
+# ── Override RentCast coordinates with building footprint geometry ────────────
+# RentCast geocoding is approximate. Our owners data uses building centroids
+# from Sarasota County GIS footprints — far more accurate. Match on house
+# number + street name and substitute when found.
+owners_sf <- readRDS("./data/owners.rds")
+
+# Extract house number from owners location field (e.g. "878   CHALMERS DR, Venice FL")
+owners_coords <- owners_sf |>
+  mutate(
+    house_num  = trimws(sub("^(\\d+).*", "\\1", location)),
+    street_raw = trimws(sub("^\\d+\\s+(.*),.*$", "\\1", location)),
+    match_key  = paste(house_num, toupper(street_raw))
+  ) |>
+  select(match_key, geom) |>
+  distinct(match_key, .keep_all = TRUE)
+
+# Extract house number + street from RentCast address
+# e.g. "878 Chalmers Dr, Unit 878, Venice, FL 34293" -> "878 CHALMERS DR"
+listings <- listings |>
+  mutate(
+    house_num  = sub("^(\\d+)\\s.*", "\\1", addressLine1),
+    street_raw = gsub("[^A-Za-z ]", "", sub("^\\d+\\s+(\\S+\\s+\\S+).*", "\\1", addressLine1)),
+    match_key  = paste(house_num, toupper(trimws(street_raw)))
+  )
+
+matched <- merge(listings, owners_coords, by = "match_key", all.x = TRUE)
+
+# For matched rows, replace RentCast lat/lng with footprint centroid coords
+has_geom <- !is.na(matched$geom)
+if (any(has_geom)) {
+  coords <- sf::st_coordinates(sf::st_as_sf(matched[has_geom, ], sf_column_name = "geom"))
+  matched$longitude[has_geom] <- coords[, "X"]
+  matched$latitude[has_geom]  <- coords[, "Y"]
+  cat("Coordinates corrected from building footprints:", sum(has_geom), "listing(s)\n")
+}
+
+listings <- matched
+
 # ── Select and clean columns ──────────────────────────────────────────────────
 listings <- listings |>
   transmute(
-    listed_date   = as.Date(listedDate),
-    address       = formattedAddress,
-    sqft          = as.numeric(squareFootage),
-    price         = as.numeric(price),
+    listed_date    = as.Date(listedDate),
+    address        = formattedAddress,
+    sqft           = as.numeric(squareFootage),
+    price          = as.numeric(price),
     price_per_sqft = round(price / sqft, 0),
     latitude,
     longitude
diff --git a/data/listings.rds b/data/listings.rds
index 23e5cbdccbde5ffbcd8d440dbe8c8beea9d0929c..e97d3c47ad4ceb617d4a3870c55eb6cac3727fc7 100644
GIT binary patch
literal 693
zcmV;m0!sZKiwFP!000001B>8dU|?WoU}0fqU}gm}8CXL@+;lB~V!}WUJCNoB(%cSL
zgc?Ay2wK1aLh~Mg&}<A~`to0h{N?vh{u3zwK9oKR6+Z;#Gcd3)0SQJ176wj`<;gjT
z#l=8|0Foe!OJYeXkp2f1=Z1=FTbNrYIA<j0<faxCE4UQtD1_!^mM8#4brix<^D>iD
zbrjru6pT%bERCUBbSw>w6dY4hGV{_)^79m&OQ3oT4UBQ>(XccyRtQNfDoM;kSYl*`
zQ@yUCp{YW6ett?)QD#bdDv}-pL!6p4EzHd^EHN}N!l^;Q(!f-~DX|C`V3lBh;569S
zM8P*ducRoy7|AV`cnmf)A;w^!H!ZPx6JJ;=TLQx-xSSaCf$lKEua1EMk~HV6fh4IZ
zkObv0%K<{qJOQESC_u$w@*NO*225N7l@Fz1<~xcgZ2)l`dDerIx}$Jg0fdH_<0!<K
z0H$H;#1$Vv<SkKYQNB44I_3bf`3`B2oZyfSbzeF}+93n#j})kU22?!z0ff$k>1#k0
z&q+XWzk{{#vL9-1R|Dz)K34_(fplRAgHfo1wb0z`o&9}4e)g{T3yXmC`%kj0#z5NM
z^5Ujv4%Q-K9J`<H1gdMkeC3rz{Rfbd>_BxQLU;65M8M2H5HejuG&VjQNIU*$(~Aev
zF)|4q*MW4Vv8ME9AniQgM$QvRv+R!AGZRQ_H!pk13#7TkIz@DVboLJKJ?t>^fqd@h
zzW6ujMMGX<ZYr#3V228F=VTU_q^88DfJz965PM=uN>M7X5P=pR#f51lFe3_zfGG&d
z;RAEx3sQ^Xk(6-cB$i~Bl%~K`apvUbr9%Wzox@p_U#<sp5-$)lfnwqR|Nmbh3K{=1
bpq64>KsP4pr4<3S0D1oa7|0X5cm)6e_--+8

literal 637
zcmV-@0)qV?iwFP!000001C5lyZ_-d4#~-5;8Nvu`>cPWPV=|R3Ed`<3rMQKYUKHjI
zGf>w8b*06&%i?V(qc@YysAP%%fM)8&D={(AqnU|E7w5&q#ETw`armR}Ra%v02l({+
zeZSwYzxR0j=0^}hLBudY6dZ;}G+ZtfC-dMXL4<%YU_^YESnaL$85aqo9l`_1;npv*
zzV(&rPgH-Pd`112jz)+Xg0BE#P`LN<MNP6Sc!+yUW>&Hl;4bwe)K3X)8W&_))$6vQ
z;~6`J%erb~j^$H$PSMpxC54L@FrP^aJk2?q&*kw<L(-axY2jIurhq#zWkO(iT#`&%
z()$VNa|7ciIWCK@7>0b!ROPzj$zi#HNvHF<+&?ej*z~{zEU;O8UNYgpZaP~ZSeVb?
zi-vBShUMu+*rPC)*-v5Ertsfw?sfbp1vs0_EBmPreWdq}Lx^0>b#jr!jmLy-k?`Sb
z>SOBDxb|(r2d+Qi*OYGk!l6y)iWkOCJaw47)gmNy3W-3~w|?aKHdz<^9FJ`fE_A&0
z#X9Y)M*b0Eoh*wDx(}K94eBp#6RK|fs^8bDzW&AO<j%*n&?a#G#q&F5VC@Hg^^`cB
zcy##Lv1VT<f6l+Xqyuwr+Mhl8Pn^5%(5>Go-4~Nzile~sCvklTIL91aTL!A%?pCuN
zJ$n6p%mAKQiryRZXgAK)4*wnvy3|zM(Ga3$L{lwWkt=d<LJ%t?$+D@y5TV1vx>2*;
z5-k&M5EU^;R9cEz@q`@IBwMwYWj89U8G4-<{&vEqu`=bh6ot=F?^oz{yWdG*po{!d
XYy^6grfMdnfZF{H0V(;Gcm)6el`cZ`