add full info to exported geocoded file + map and table to check accuracy

scraping/geocoding-viviendas-turisticas-comunitat-valenciana.R

@@ -19,7 +19,7 @@ calles_valencia = read.csv("data/original/vias-valencia.csv") %>%
   mutate(nombre_es = as.factor(nombre_es))
 
 # Load dataset and manipulate addresses.
-df = read.csv("data/original/190302_viviendas-turisticas-comunidad-valenciana_valencia.csv") %>%
+df = read.csv("data/output/190302_viviendas-turisticas-comunidad-valenciana_valencia_por-geocodificar.csv") %>%
   separate(Address, c("nombre_es_raw", "num"), extra = "merge",
            sep = ", nº", remove = FALSE) %>%
   separate(num, c("num", "puerta"), extra = "merge", sep = ", ", remove = FALSE) %>%
@@ -40,7 +40,6 @@ record.linkage.names = function(names,
   return(names)
 }
 
-
 df = df %>%
   mutate(nombre_es = record.linkage.names(nombre_es_raw)) %>%
   left_join(calles_valencia)
@@ -53,14 +52,41 @@ df2 = df %>%
   mutate(full_address_ca = paste(tipovia_ca, nomoficial, num, Municipio,
                                  sep = ", "))
 
-geocoded.df = photon::geocode(head(df2$full_address_ca), limit = 1,
+geocoded.df = photon::geocode(df2$full_address_ca, limit = 1,
                               # lang = "es",
                               key = "highway",
                               locbias = c(-0.3766, 39.4665))
 
+# TODO explore if extending search to three makes it possible later to select the one solution in Valencia City
+# geocoded.df3 = photon::geocode(head(df2$full_address_ca), limit = 3,
+#                               # lang = "es",
+#                               key = "highway",
+#                               locbias = c(-0.3766, 39.4665))
+
+# Check locations ----------
+table(geocoded.df$country)
+table(geocoded.df$city)
+
+# plot results in a map
+ggplot() +
+  # geom_polygon(data = municipios,
+  #              aes(x = long, y = lat, group = group),
+  #              color = "grey", fill="white", size = 0.1) +
+  geom_polygon(data = barrios,
+               aes(x = long, y = lat, group = group),
+               color = "grey", fill="white", size = 0.1) +
+  geom_point(data= geocoded.df,
+             aes(x=lon, y=lat),alpha=1,size = 0.1) +
+  geom_text(data= geocoded.df,
+             aes(x=lon, y=lat, label=name),alpha=1,size = 0.1)
+
+# Combine geocoded dataframe with original one.---------------
+geocoded.df$row.id <- rownames(geocoded.df)
+df$row.id <- rownames(df)
+
+# df.combined <- df2 %>%
+#   left_join(geocoded.df, by = c("full_address_ca" = "location"))
 
-# Combine geocoded dataframe with original one.
-df.combined = df2 %>%
-  left_join(geocoded.df, by = c("full_address_ca" = "location"))
+df.combined <- full_join(df, geocoded.df, by = "row.id")
 
-write.csv(df.combined, file = "data/output/filename.csv")
+write.csv(df.combined, file = "data/output/190302_viviendas-turisticas-comunidad-valenciana_valencia_por-geocodificar_geocoded-photon2.csv", row.names = FALSE)
\ No newline at end of file