updating figures for tentative printout

dsl/final_bivariate.R

@@ -4,20 +4,90 @@ library(dplyr)
 dsl_csv <-"~/dsl/111725_DSL_frame.csv"
 dsl_df <- read.csv(dsl_csv, header = TRUE) 
 
+dsl_df <- dsl_df |>
+  mutate(priority = factor(priority, 
+                           levels = rev(c("Unbreak Now!", "High", "Medium", "Low", "Lowest", "Needs Triage"))))
+
+data_summary <- dsl_df %>%
+  group_by(week_index, priority, source) %>%
+  summarise(count = n(), .groups = 'drop') |>
+  group_by(week_index, source) %>%
+  mutate(proportion = count / sum(count)) %>%
+  ungroup()
+
+library(ggdist)
+
+data_summary|>
+  filter(priority == "Needs Triage" | 
+           priority == "Unbreak Now!" |
+           priority == "High") |>
+  ggplot(aes(x = week_index, y = proportion, color = priority, fill=priority, group = priority)) +
+  geom_smooth()+
+  scale_color_viridis_d(option='turbo') + 
+  scale_fill_viridis_d(option='turbo') + 
+  facet_grid(source ~ ., scales = "free_y") + 
+  geom_point() +
+  labs(title = "Triage priority proportions for new tasks by week created",
+       x = "Weeks from feature deployment",
+       y = "% of items tagged",
+       color = "Priority Tag") +
+  theme_minimal()
+
+dsl_df |>
+  filter(priority == "Needs Triage" | 
+           priority == "Unbreak Now!" |
+           priority == "High") |>
+  filter(week_index >= -26) |>
+  ggplot(
+    aes(
+      x=as.factor(week_index),
+      y= TTR/168, 
+      color=priority,
+      fill=priority
+    )
+  ) + 
+    facet_grid(source ~ .) + 
+    geom_boxplot(outlier.shape = NA) +
+    theme_minimal() + 
+    coord_cartesian(ylim = c(0, 112)) + 
+    geom_vline(xintercept =27, linetype = "dashed", color = "black", linewidth = 0.5) +
+    scale_color_viridis_d() +
+    labs(x = "Weeks from Release", y = "Time to Resolution (weeks)", title = "TTR by Task Creation Date and Triage Priority")
+
+
+dsl_df_long <- dsl_df %>%
+  pivot_longer(
+    cols = c(olmo_EP_prop, olmo_RK_prop, olmo_TSOL_prop),
+    names_to = "tag",
+    values_to = "proportion"
+  ) %>%
+  mutate(tag = gsub("olmo_|_prop", "", tag),
+         tag = case_when(
+           tag == "EP" ~ "Existent Problem",
+           tag == "RK" ~ "Record Keeping",
+           tag =="TSOL" ~ "Solutions"
+         )) 
 
 ggplot(
-  dsl_df,
+  dsl_df_long,
   aes(
-    x=as.factor(week_index),
+    x = tag,
-    y= (TTR/168), 
+    y = proportion, 
-    fill=isAuthorWMF
+    fill = isAuthorWMF,
   )
 ) + 
-  facet_grid(source ~ ., scales = "free_y") + 
+  facet_grid(source ~ .) + 
   geom_boxplot() +
   theme_minimal() + 
   scale_fill_viridis_d() +
-  labs(x = "Weeks from Release", y = "Time to Resolution (weeks)", title = "TTR by Task Creation Date")
+  labs(
+    x = "Tag", 
+    y = "% of sentences tagged", 
+    title = "Proportion of machine tags of sentence focus, by comment author affiliation",
+    color = "Is Author WMF",
+    fill = "Is Author WMF"
+  )
+
 
 weekly_summary <- dsl_df |>
   group_by(week_index, source, isAuthorWMF)|>
@@ -27,6 +97,7 @@ weekly_summary <- dsl_df |>
     author_closer_sum  = sum(author_closer == TRUE),
     median_olmo_EP_prop_adac = median(olmo_EP_prop_adac),
     median_olmo_TSOL_prop_adac = median(olmo_TSOL_prop_adac),
+    median_olmo_RK_prop_adac = median(olmo_RK_prop_adac),
     median_comments_before_resolution = median(n_comments_before)
   )
 

p2/quest/neurobiber_PCA_analysis.R

@@ -42,12 +42,12 @@ main_df |>
   )
 
 main_df |>
-  filter(ADAC=="1") |>
+  filter(ADAC == 1) |>
   ggplot(
     aes(
       x = PC4,
       y = PC3,
-      fill = isAuthorWMF
+      fill = as.factor(ADAC)
     )
   ) + 
   facet_grid(comment_type~source,
@@ -69,6 +69,7 @@ main_df |>
     x = "Casual v. Formal Updates (PC3)",
     y = "Technical-matter v. Procedural Commentary (PC4)",
   )
+#"PCs for Pre-Resolution Comments Written by Task Author (by Author Affiliation, Case, and Comment Type)"
 
 main_df |>
   filter(comment_type=="task_subcomment") |>
@@ -79,7 +80,7 @@ main_df |>
       fill = as.factor(ADAC)
     )
   ) + 
-  facet_grid(isAuthorWMF~source,
+  facet_grid(ADAC~source,
              labeller = as_labeller(c(
                "c1" = "VisualEditor (c1)",
                "c2"   = "HTTPS-as-default (c2)",
@@ -87,7 +88,7 @@ main_df |>
              ))) +
   geom_point(shape = 21, alpha=0.13, size=2) +
   scale_fill_viridis_d(
-    option = "turbo",
+    option = "inferno",
     name = "By Task Author Before Resolution",
     labels = c("No", "Yes"))+
   theme_minimal() +