Added code for last week's "World Freedom Index"

2022_02_22_world_freedom_index.Rmd

@@ -0,0 +1,268 @@
+---
+title: "TidyTemplate"
+date: 2022-02-22
+output: html_output
+editor_options: 
+  chunk_output_type: console
+---
+
+# TidyTuesday
+
+```{r setup, include=FALSE}
+
+knitr::opts_chunk$set(echo = TRUE)
+
+library(tidyverse)
+library(tidytuesdayR)
+library(scales)
+theme_set(theme_light())
+
+```
+
+# Load the weekly Data
+
+Dowload the weekly data and make available in the `tt` object.
+
+```{r Load}
+freedom <- tt_load("2022-02-22")$freedom %>%
+  janitor::clean_names() %>%
+  rename(civil_liberties = cl,
+         political_rights = pr) %>%
+  mutate(country_code = countrycode::countrycode(country, "country.name", "iso2c"))
+
+freedom %>%
+  count(year, sort = TRUE) %>%
+  arrange(desc(year))
+
+freedom %>%
+  distinct(country, region_name)
+```
+
+Look just at 2020
+
+```{r}
+summarize_freedom <- function(tbl) {
+  tbl %>%
+    summarize(n_countries = n(),
+              avg_civil_liberties = mean(civil_liberties),
+              avg_political_rights = mean(political_rights),
+              pct_free = mean(status == "F"),
+              .groups = "drop") %>%
+    arrange(desc(n_countries))
+}
+```
+
+
+```{r}
+by_region <- freedom %>%
+  filter(year == 2020) %>%
+  group_by(region_name) %>%
+  summarize_freedom()
+
+by_region %>%
+  ggplot(aes(avg_civil_liberties, avg_political_rights)) +
+  geom_abline(color = "red") +
+  geom_point(aes(size = n_countries)) +
+  geom_text(aes(label = region_name), vjust = 1, hjust = 1) +
+  expand_limits(x = 0, y = 0, size = 0)
+
+freedom %>%
+  ggplot(aes(avg_civil_liberties, avg_political_rights)) +
+  geom_point()
+```
+
+```{r}
+freedom %>%
+  filter(year == 2020) %>%
+  ggplot(aes(civil_liberties, political_rights)) +
+  geom_abline(color = "red") +
+  geom_jitter(height = .2, width = .2) +
+  # geom_text(aes(label = region_name), vjust = 1, hjust = 1) +
+  expand_limits(x = 0, y = 0, size = 0)
+
+freedom %>%
+  summarize(sd(civil_liberties),
+            sd(political_rights))
+
+freedom %>%
+  filter(year == 2020) %>%
+  gather(metric, value, civil_liberties, political_rights) %>%
+  mutate(metric = str_to_title(str_replace_all(metric, "_", " ")),
+         region_name = fct_reorder(region_name, value)) %>%
+  count(region_name, metric, value) %>%
+  ggplot(aes(value, n)) +
+  geom_col() +
+  facet_grid(region_name ~ metric) +
+  labs(x = "World Freedom Index rating",
+       y = "# of countries",
+       title = "Distribution of World Freedom Index by region in 2020")
+```
+
+civil_liberties is a scale of 1-7
+
+```{r}
+freedom_gathered <- freedom %>%
+  gather(metric, value, civil_liberties, political_rights) %>%
+  mutate(metric = str_to_title(str_replace_all(metric, "_", " ")),
+         region_name = fct_reorder(region_name, value))
+
+overall <- freedom_gathered %>%
+  group_by(year, metric) %>%
+  summarize(avg_rating = mean(value)) 
+
+freedom_gathered %>%
+  group_by(year, region_name, metric) %>%
+  summarize(avg_rating = mean(value)) %>%
+  ggplot(aes(year, avg_rating)) +
+  geom_line(aes(color = region_name)) +
+  geom_line(data = overall, size = 3) +
+  facet_wrap(~ metric) +
+  expand_limits(y = 1) +
+  scale_y_reverse(breaks = seq(1, 7)) +
+  scale_color_discrete(guide = guide_legend(reverse = TRUE)) +
+  labs(x = "Year",
+       y = "World Freedom Index rating",
+       title = "World Freedom Index rating over time by region",
+       color = "Region",
+       subtitle = "Black line shows overall trend")
+```
+
+```{r}
+freedom %>%
+  ggplot(aes(civil_liberties)) +
+  geom_histogram()
+```
+
+### Worldbank data
+
+```{r}
+library(WDI)
+library(countrycode)
+
+gdp_percap <- WDI(indicator = "NY.GDP.PCAP.CD",
+                  extra = TRUE,
+                  start = 1995,
+                  end = 2020) %>%
+  as_tibble()
+
+freedom_joined <- freedom_gathered %>%
+  inner_join(gdp_percap, by = c(country_code = "iso2c", "year"),
+             suffix = c("", "_wdi")) %>%
+  mutate(income = fct_relevel(income, c("Low income", "Lower middle income", "Upper middle income")))
+
+freedom_joined %>%
+  filter(income != "Not classified") %>%
+  group_by(metric, income, year) %>%
+  summarize(avg_rating = mean(value)) %>%
+  ggplot(aes(year, avg_rating)) +
+  geom_line(aes(color = income)) +
+  geom_line(data = overall, size = 3) +
+  facet_wrap(~ metric) +
+  expand_limits(y = 1) +
+  scale_y_reverse(breaks = seq(1, 7)) +
+  scale_color_discrete(guide = guide_legend(reverse = TRUE)) +
+  labs(x = "Year",
+       y = "World Freedom Index rating",
+       title = "World Freedom Index rating over time by region",
+       color = "Worldbank Income",
+       subtitle = "Black line shows overall trend")
+
+freedom_joined %>%
+  filter(year == 2020) %>%
+  ggplot(aes(NY.GDP.PCAP.CD, value)) +
+  geom_point() +
+  geom_jitter(height = .2, width = 0) +
+  facet_wrap(~ metric) +
+  scale_x_log10()
+```
+
+```{r}
+library(broom)
+
+civil_liberties_2020 <- freedom_joined %>%
+  filter(metric == "Civil Liberties",
+         year == 2020,
+         !is.na(NY.GDP.PCAP.CD))
+
+lin_mod <- civil_liberties_2020 %>%
+  lm(value ~ region_name + log2(NY.GDP.PCAP.CD), data = .)
+
+library(ggrepel)
+lin_mod %>%
+  augment(data = civil_liberties_2020) %>%
+  select(country, NY.GDP.PCAP.CD, region_name, income, value, .fitted, .resid) %>%
+  arrange(desc(abs(.resid))) %>%
+  head(20) %>%
+  ggplot(aes(.fitted, value)) +
+  geom_point() +
+  geom_text_repel(aes(label = country)) +
+  geom_abline(color = "red") +
+  labs(x = "Expected freedom index based on region + income",
+       y = "Actual freedom index",
+       title = "What are the largest outliers?") +
+  expand_limits(x = 1, y = 1)
+```
+
+```{r}
+library(fuzzyjoin)
+
+freedom_2020 <- freedom_joined %>%
+  filter(year == 2020)
+
+world_map_freedom_2020 <- map_data("world") %>%
+  as_tibble() %>%
+  regex_left_join(maps::iso3166, c(region = "mapname")) %>%
+  left_join(freedom_2020 %>% select(-region), by = c(a2 = "country_code")) %>%
+  filter(region != "Antarctica")
+
+world_map_freedom_2020 %>%
+  filter(metric == "Civil Liberties") %>%
+  ggplot(aes(long, lat, group = group)) +
+  geom_polygon(aes(fill = value)) +
+  coord_map(xlim = c(-180, 180)) +
+  scale_fill_gradient2(low = "blue",
+                       high = "red",
+                       midpoint = 3.5,
+                       guide = guide_legend(reverse = TRUE)) +
+  ggthemes::theme_map() +
+  labs(fill = "Civil Liberties Rating",
+       title = "World Freedom Index: Civil Liberties",
+       subtitle = "In 2020")
+```
+
+```{r}
+library(gganimate)
+
+world_map_freedom <- map_data("world") %>%
+  as_tibble() %>%
+  regex_left_join(maps::iso3166, c(region = "mapname")) %>%
+  left_join(freedom_joined %>% select(-region), by = c(a2 = "country_code")) %>%
+  filter(region != "Antarctica")
+
+world_map_freedom %>%
+  filter(metric == "Civil Liberties") %>%
+  ggplot(aes(long, lat, group = group)) +
+  geom_polygon(aes(fill = value)) +
+  coord_map(xlim = c(-180, 180)) +
+  scale_fill_gradient2(low = "blue",
+                       high = "red",
+                       midpoint = 3.5,
+                       guide = guide_legend(reverse = TRUE)) +
+  ggthemes::theme_map() +
+  transition_manual(year) +
+  labs(fill = "Civil Liberties Rating",
+       title = "World Freedom Index: Civil Liberties ({ current_frame })")
+```
+
+
+
+
+```{r}
+freedom %>%
+  distinct(country, country_code) %>%
+  View()
+
+gdp_percap %>%
+  filter(str_detect(country, "Iran"))
+```
+