Add info from outlier reason 2 and clean up graphs (#69)

* Add maing changes

* Remove hard coded triad refs

* Remove quotes

* Incorporate feedback

reports/performance/_outliers.qmd

@@ -18,27 +18,28 @@ exclusively on these price-based factors that determine outlier status.
 ### Count by Type
 
 ```{r _outliers_type_breakdown}
-# Make new column for a few cells of the outliers section
+# Subset important outlier information
 training_data <- training_data %>%
-  mutate(price_outlier_reason = case_when(
-    # Also grabs outlier reason Non-livable computed in 00-ingest.R
-    str_detect(sv_outlier_reason1, regex("price|Non-livable", ignore_case = TRUE)) ~ sv_outlier_reason1,
-    str_detect(sv_outlier_reason2, regex("price", ignore_case = TRUE)) ~ sv_outlier_reason2,
-    str_detect(sv_outlier_reason3, regex("price", ignore_case = TRUE)) ~ sv_outlier_reason3,
-    TRUE ~ NA_character_
-  ))
+  mutate(
+    outlier_reasons_to_graph = case_when(
+      sv_is_outlier & !is.na(sv_outlier_reason2) ~
+        paste0(sv_outlier_reason1, " (", sv_outlier_reason2, ")"),
+      sv_is_outlier & is.na(sv_outlier_reason2) ~ sv_outlier_reason1,
+      TRUE ~ NA_character_
+    )
+  )
 
 # Determine the axis limit
 y_lim_axis_outlier_breakdown <- training_data %>%
   filter(sv_is_outlier) %>%
-  count(price_outlier_reason) %>%
+  count(outlier_reasons_to_graph) %>%
   summarise(max_value = max(n)) %>%
   pull(max_value)
 
 training_data %>%
   filter(sv_is_outlier) %>%
-  count(price_outlier_reason) %>%
-  ggplot(aes(x = reorder(price_outlier_reason, -n), y = n)) +
+  count(outlier_reasons_to_graph) %>%
+  ggplot(aes(x = reorder(outlier_reasons_to_graph, -n), y = n)) +
   geom_bar(stat = "identity") +
   geom_text(aes(label = comma(n)), vjust = -0.5) +
   ylim(0, 1.03 * y_lim_axis_outlier_breakdown) +
@@ -57,19 +58,19 @@ training_data %>%
 ```{r _outliers_type_breakdown_table}
 training_data %>%
   filter(sv_is_outlier) %>%
-  group_by(meta_year, price_outlier_reason) %>%
+  group_by(meta_year, outlier_reasons_to_graph) %>%
   summarise(n = n()) %>%
   rename(Year = meta_year) %>%
-  pivot_wider(id_cols = Year, names_from = price_outlier_reason, values_from = n) %>%
+  pivot_wider(id_cols = Year, names_from = outlier_reasons_to_graph, values_from = n) %>%
   kable() %>%
   kable_styling("striped")
 
 training_data %>%
   filter(sv_is_outlier) %>%
-  group_by(meta_year, price_outlier_reason) %>%
+  group_by(meta_year, outlier_reasons_to_graph) %>%
   summarise(n = n(), .groups = "drop") %>%
   rename(Year = meta_year) %>%
-  pivot_wider(id_cols = Year, names_from = price_outlier_reason, values_from = n) %>%
+  pivot_wider(id_cols = Year, names_from = outlier_reasons_to_graph, values_from = n) %>%
   kable() %>%
   kable_styling("striped")
 ```
@@ -262,6 +263,10 @@ outliers_ratio_comparison <- training_data %>%
   ) %>%
   distinct(meta_township_name, percent, above_below, triad)
 
+axis_limit_outlier_ratio_comparison <- outliers_ratio_comparison %>%
+  slice_max(percent, n = 1) %>%
+  pull(percent)
+
 outliers_ratio_comparison %>%
   ggplot(aes(x = reorder(meta_township_name, percent), y = percent)) +
   labs(
@@ -270,7 +275,7 @@ outliers_ratio_comparison %>%
   geom_bar(stat = "identity", aes(fill = above_below)) +
   coord_flip() +
   geom_text(aes(label = round(percent, 2)), size = 3.2, hjust = -0.2) +
-  scale_y_continuous(limits = c(0, 1.5)) +
+  scale_y_continuous(limits = c(0, 1.1 * axis_limit_outlier_ratio_comparison)) +
   theme_minimal() +
   theme(
     axis.title.y = element_blank(),
@@ -301,7 +306,7 @@ outliers_table_township_summary <- outliers_table_township_summary %>%
     `Med. Sale Price` = median(meta_sale_price, na.rm = TRUE),
     `Max. Sale Price` = max(meta_sale_price, na.rm = TRUE),
     Count = n(),
-    .by = c(price_outlier_reason, meta_township_name)
+    .by = c(outlier_reasons_to_graph, meta_township_name)
   ) %>%
   left_join(
     outliers_table_township_summary %>%
@@ -315,7 +320,7 @@ outliers_table_township_summary <- outliers_table_township_summary %>%
   mutate(across(contains("Sale"), dollar)) %>%
   relocate(meta_township_name) %>%
   dplyr::rename(
-    "Outlier Type" = price_outlier_reason,
+    "Outlier Type" = outlier_reasons_to_graph,
     "Township Name" = meta_township_name
   ) %>%
   arrange(`Township Name`, desc(Count))
@@ -346,7 +351,7 @@ outliers_table_class_summary <- outliers_table_class_summary %>%
     `Med. Sale Price` = median(meta_sale_price, na.rm = TRUE),
     `Max. Sale Price` = max(meta_sale_price, na.rm = TRUE),
     Count = n(),
-    .by = c(price_outlier_reason, meta_class)
+    .by = c(outlier_reasons_to_graph, meta_class)
   ) %>%
   left_join(
     outliers_table_class_summary %>%
@@ -360,7 +365,7 @@ outliers_table_class_summary <- outliers_table_class_summary %>%
   mutate(across(contains("Sale"), dollar)) %>%
   relocate(meta_class) %>%
   dplyr::rename(
-    "Outlier Type" = price_outlier_reason,
+    "Outlier Type" = outlier_reasons_to_graph,
     "Class" = meta_class
   ) %>%
   arrange(`Class`, desc(Count))
@@ -431,7 +436,7 @@ outlier_decile_breakout <- function(data, dec) {
     arrange(meta_sale_price) %>%
     mutate(decile = ntile(meta_sale_price, 10)) %>%
     filter(decile == dec & sv_is_outlier) %>%
-    group_by(price_outlier_reason) %>%
+    group_by(outlier_reasons_to_graph) %>%
     summarise(count = n()) %>%
     ungroup() %>%
     slice_max(count, n = 1) %>%
@@ -441,8 +446,8 @@ outlier_decile_breakout <- function(data, dec) {
     arrange(meta_sale_price) %>%
     mutate(decile = ntile(meta_sale_price, 10)) %>%
     filter(decile == dec & sv_is_outlier) %>%
-    summarise(count = n(), .by = price_outlier_reason) %>%
-    ggplot(aes(x = reorder(price_outlier_reason, -count), y = count)) +
+    summarise(count = n(), .by = outlier_reasons_to_graph) %>%
+    ggplot(aes(x = reorder(outlier_reasons_to_graph, -count), y = count)) +
     labs(
       y = "Number of Sales",
       x = "Outlier Types"