Add more data and update plots.

Some plots modified a bit more than others. The stacked plot is a mess because of the different license classes, and the color-blind-friendly color scale breaks when there are too many colors.

README.md

@@ -4,6 +4,10 @@ Data and code for analysis at https://mastodon.radio/@W1CDN/111723969200676208.
 
 I'll get a blog post up at https://w1cdn.net at some point.
 
+This is a work in progress. It is not intended to be immediately useful for
+detailed analysis, but to act as a guide for further investigation. As we figure out
+how to slice up and caveat data, it will get more organized.
+
 # Resources (not all integrated yet)
 - Various items in the [issue queue](https://amiok.net/gitea/W1CDN/ham-radio-licenses/issues)
 - Post from KE9V that started this: https://mastodon.sdf.org/@jdavis/111720350879036543

bin/data-plots.R

@@ -4,6 +4,8 @@ library(tidyr)
 library(dplyr)
 library(forcats)
 library(ggthemes)
+library(plotly)
+library(htmlwidgets)
 
 # Read in data from Wayback Machine archive of http://www.arrl.org/fcc-license-counts
 
@@ -16,39 +18,18 @@ unique(d_raw$Date)
 d <- d_raw %>% 
         mutate(Date = as.Date(Date)) %>% 
         # Drop Tech alone and leave Tech and Tech Plus
-        select(-c(Tech, Tech.Plus)) %>% 
+        select(-c(Tech, Tech.Plus))
-        mutate(across(c(3:8), as.numeric))
         
-
 d_total <- d %>% filter(State.Territory == "TOTAL")
                 
 d_total_long <- d_total %>% 
-                pivot_longer(cols = 3:8,
+                pivot_longer(cols = !c("Date", "State.Territory", "source_name", "source_detail"),
                              names_to = "lclass",
                              values_to = "count") %>% 
                 # only keep rows with data so plots look right
                 filter(!is.na(count))
                 
 #### Plots ####
-##### Total over time #####
-ggplot(data = d_total,
-       aes(x = Date,
-           y = Total,
-           color = source_name)) +
-    geom_line() +
-    geom_point(size = 0.3) +
-    scale_x_date(date_breaks = "10 years",
-                 date_minor_breaks = "1 year",
-                 date_labels = "%Y") +
-    scale_y_continuous(labels = scales::comma) +
-    theme_bw() +
-    labs(title = "US Amateur Licenses",
-         subtitle = "ARRL, AH0A",
-         caption = "w1cdn.net",
-         color = "Source") +
-    theme(legend.position="bottom")
-
-ggsave("plots/total-over-time.png", width = 6, height = 4)
 
 ##### Total over time, y = 0 #####
 ggplot(data = d_total,
@@ -64,13 +45,34 @@ ggplot(data = d_total,
                        limits = c(0, NA)) +
     theme_bw() +
     labs(title = "US Amateur Licenses",
-         subtitle = "ARRL, AH0A",
          caption = "w1cdn.net",
          color = "Source") +
     theme(legend.position="bottom")
 
 ggsave("plots/total-over-time-y.png", width = 6, height = 4)
 
+p <- ggplotly()
+htmlwidgets::saveWidget(as_widget(p), selfcontained = TRUE, file = "plots/total-over-time-y.html")
+
+##### Total over time, since 2000 #####
+ggplot(data = d_total %>% filter(Date >= as.Date("2000-01-01")),
+       aes(x = Date,
+           y = Total,
+           color = source_name)) +
+    geom_line() +
+    geom_point(size = 0.3) +
+    scale_x_date(date_breaks = "1 years",
+                 date_minor_breaks = "3 months",
+                 date_labels = "%Y") +
+    scale_y_continuous(labels = scales::comma) +
+    theme_bw() +
+    labs(title = "US Amateur Licenses since 2000",
+         caption = "w1cdn.net",
+         color = "Source") +
+    theme(legend.position="bottom")
+
+ggsave("plots/total-over-time.png", width = 6, height = 4)
+
 ##### By license class #####
 ggplot(data = d_total_long %>% filter(lclass != "Total"),
        aes(x = Date,
@@ -85,7 +87,6 @@ ggplot(data = d_total_long %>% filter(lclass != "Total"),
     scale_color_colorblind() +
     theme_bw() +
     labs(title = "US Amateur Licenses by Class",
-         subtitle = "ARRL, AH0A",
          y = "Count",
          color = "Class",
          caption = "w1cdn.net") +
@@ -106,7 +107,6 @@ ggplot(data = d_total_long %>% filter(lclass != "Total"),
     scale_y_continuous(labels = scales::comma) +
     theme_bw() +
     labs(title = "US Amateur Licenses by Class",
-         subtitle = "ARRL, AH0A",
          fill = "Class",
          caption = "w1cdn.net") +
     theme(legend.position="bottom")