SC/ST population distribution

library(indiacensus)
library(dplyr)
library(ggplot2)
library(ggrepel)
library(biscale)
library(patchwork)

Scheduled Castes (SC) and Scheduled Tribes (ST) are constitutionally recognized groups in India. Their geographic distribution reflects historical settlement patterns and regional diversity.

sc_st_2011 <- census_2011_pca |>
  mutate(
    sc_pct = 100 * sc_population / population_total,
    st_pct = 100 * st_population / population_total
  ) |>
  attach_geometry(year = 2011, geography = "district")

sc_st_1971 <- census_1971 |>
  filter(geography == "district") |>
  mutate(
    sc_pct = 100 * sc_population_total / population_total,
    st_pct = 100 * st_population_total / population_total
  ) |>
  attach_geometry(year = 1971, geography = "district")

Scheduled Caste population

SC: 2011

plot_map(
  sc_st_2011,
  fill_var = "sc_pct",
  title = "Scheduled Caste population (2011)",
  subtitle = "Percentage of district population",
  legend_title = "SC %",
  palette = "oranges",
  show_state_boundaries = TRUE
)

SC: 1971 vs 2011

compare_maps(
  list(
    "1971" = sc_st_1971,
    "2011" = sc_st_2011
  ),
  fill_var = "sc_pct",
  title = "SC population: 1971 vs 2011",
  legend_title = "SC %",
  palette = "oranges"
)

Scheduled Tribe population

ST: 2011

plot_map(
  sc_st_2011,
  fill_var = "st_pct",
  title = "Scheduled Tribe population (2011)",
  subtitle = "Percentage of district population",
  legend_title = "ST %",
  palette = "greens",
  show_state_boundaries = TRUE
)

ST: 1971 vs 2011

compare_maps(
  list(
    "1971" = sc_st_1971,
    "2011" = sc_st_2011
  ),
  fill_var = "st_pct",
  title = "ST population: 1971 vs 2011",
  legend_title = "ST %",
  palette = "greens"
)

State-level summary (2011)

state_summary <- census_2011_pca |>
  group_by(state_name_harmonized) |>
  summarise(
    population = sum(population_total),
    sc_population = sum(sc_population),
    st_population = sum(st_population)
  ) |>
  mutate(
    sc_pct = round(100 * sc_population / population, 1),
    st_pct = round(100 * st_population / population, 1)
  ) |>
  arrange(desc(sc_pct + st_pct))

cat("States with highest SC population %:\n")
#> States with highest SC population %:
state_summary |>
  arrange(desc(sc_pct)) |>
  select(state_name_harmonized, sc_pct) |>
  head(5)
#> # A tibble: 5 × 2
#>   state_name_harmonized sc_pct
#>   <chr>                  <dbl>
#> 1 Punjab                  28.9
#> 2 Himachal Pradesh        24.7
#> 3 West Bengal             23  
#> 4 Uttar Pradesh           21.1
#> 5 Haryana                 19.3

cat("\nStates with highest ST population %:\n")
#> 
#> States with highest ST population %:
state_summary |>
  arrange(desc(st_pct)) |>
  select(state_name_harmonized, st_pct) |>
  head(5)
#> # A tibble: 5 × 2
#>   state_name_harmonized st_pct
#>   <chr>                  <dbl>
#> 1 Lakshadweep             94.5
#> 2 Mizoram                 94.5
#> 3 Nagaland                89.1
#> 4 Meghalaya               85.9
#> 5 Arunachal Pradesh       64.2

Bivariate choropleth: SC and ST together

This map uses a bivariate color scheme to show both SC and ST percentages simultaneously. The legend shows how colors combine: one axis represents SC concentration, the other ST concentration.

boundaries <- get_census_boundaries(2011, "district")

bivariate_data <- sc_st_2011 |>
  mutate(
    sc_pct = ifelse(is.na(sc_pct), 0, sc_pct),
    st_pct = ifelse(is.na(st_pct), 0, st_pct)
  ) |>
  bi_class(x = sc_pct, y = st_pct, style = "quantile", dim = 3)

map <- ggplot() +
  geom_sf(data = boundaries, fill = "grey85", color = "grey70", linewidth = 0.05) +
  geom_sf(data = bivariate_data, aes(fill = bi_class), color = "grey50", linewidth = 0.1, show.legend = FALSE) +
  bi_scale_fill(pal = "BlueOr", dim = 3, na.value = "grey85") +
  labs(
    title = "SC and ST population distribution (2011)",
    subtitle = "Bivariate choropleth showing joint distribution"
  ) +
  bi_theme()

legend <- bi_legend(pal = "BlueOr", dim = 3, xlab = "SC % ", ylab = "ST % ", size = 8)

map + inset_element(legend, left = 0.7, bottom = 0.05, right = 0.95, top = 0.3)

SC vs ST scatter plot

scatter_data <- sf::st_drop_geometry(sc_st_2011) |>
  mutate(
    is_extreme = sc_pct > quantile(sc_pct, 0.95, na.rm = TRUE) |
      st_pct > quantile(st_pct, 0.95, na.rm = TRUE),
    label = ifelse(is_extreme, name, NA)
  )

ggplot(scatter_data, aes(sc_pct, st_pct)) +
  geom_point(aes(color = is_extreme), alpha = 0.5) +
  geom_smooth(method = "loess", se = TRUE, color = "steelblue", fill = "lightblue") +
  geom_text_repel(
    aes(label = label),
    size = 3,
    max.overlaps = 15,
    na.rm = TRUE
  ) +
  scale_color_manual(values = c("grey50", "red"), guide = "none") +
  coord_fixed(ratio = 1) +
  labs(
    x = "SC population (%)",
    y = "ST population (%)",
    title = "SC vs ST concentration by district (2011)",
    subtitle = "Districts tend to have either high SC or high ST, rarely both"
  ) +
  theme_minimal()