region_tbl <- pre %>%
  filter(!is.na(home_region)) %>%
  count(home_region, .drop = FALSE, name = "N") %>%
  mutate(`%` = sprintf("%.1f%%", 100 * N / sum(N))) %>%
  arrange(home_region)

# Total row at the bottom for quick sanity check.
region_tbl <- bind_rows(
  region_tbl,
  tibble(home_region = "Total",
         N           = sum(region_tbl$N),
         `%`         = "100.0%")
)

kable(region_tbl,
      col.names = c("Home region", "N", "%"),
      caption = "AEP students by home region (survey-reported)",
      align = c("l", "r", "r")) %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped", "hover")) %>%
  row_spec(nrow(region_tbl), bold = TRUE)

region_state_tbl <- pre %>%
  filter(!is.na(home_region), !is.na(state_travel)) %>%
  count(home_region, state_travel, name = "N") %>%
  group_by(home_region) %>%
  mutate(`%` = sprintf("%.1f%%", 100 * N / sum(N))) %>%
  arrange(home_region, desc(N), state_travel) %>%
  ungroup()

kable(region_state_tbl,
      col.names = c("Home region", "Travel state", "N", "% of region"),
      caption = "Travel destinations by home region (survey-reported)",
      align = c("l", "l", "r", "r")) %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped", "hover")) %>%
  collapse_rows(columns = 1, valign = "top")

# Knobs -- tweak to retune. TIER_CUTOFFS define which destinations land
# in each tier; TIER_WEIGHTS are the combined probabilities (achieved in
# Qualtrics via slot duplication: 5/3/2 of 10 evenly-randomized slots).
TIER_CUTOFFS   <- c(7, 3)                # Tier 1 / Tier 2 lower bounds (% of region)
TIER_WEIGHTS   <- c(0.50, 0.30, 0.20)    # Tier 1 / 2 / 3 combined weights
region_targets <- c("South" = 150, "Northeast" = 150, "West Coast" = 150,
                    "Midwest" = 100, "Mountain West" = 150)

# Per-state tier + projected counts. Within each (region, tier) the
# combined tier weight is split equally across the tier's states.
proj <- region_state_tbl %>%
  group_by(home_region) %>%
  mutate(pct_num = 100 * N / sum(N)) %>%
  ungroup() %>%
  mutate(Tier = case_when(
    pct_num >= TIER_CUTOFFS[1] ~ "1",
    pct_num >= TIER_CUTOFFS[2] ~ "2",
    TRUE                        ~ "3"
  )) %>%
  group_by(home_region, Tier) %>%
  mutate(n_in_tier = n()) %>%
  ungroup() %>%
  mutate(target_N     = unname(region_targets[as.character(home_region)]),
         proj_pct_num = TIER_WEIGHTS[as.integer(Tier)] / n_in_tier * 100,
         proj_N_num   = proj_pct_num / 100 * target_N)

# Per-region TOTAL row (sums of N + projected counts; both should hit 100% / target).
totals <- proj %>%
  group_by(home_region) %>%
  summarise(state_travel = "TOTAL",
            Tier         = "",
            N            = sum(N),
            pct_num      = sum(pct_num),
            proj_N_num   = sum(proj_N_num),
            proj_pct_num = sum(proj_pct_num),
            .groups      = "drop")

# Stack states + totals; TOTAL pinned to the bottom of each region.
projection_tbl <- bind_rows(
    proj   %>% mutate(is_total = FALSE),
    totals %>% mutate(is_total = TRUE)
  ) %>%
  arrange(home_region, is_total, desc(N)) %>%
  mutate(`% of region` = sprintf("%.1f%%", pct_num),
         `Projected N` = sprintf("%.1f",   proj_N_num),
         `Projected %` = sprintf("%.1f%%", proj_pct_num)) %>%
  select(`Home region`  = home_region,
         `Travel state` = state_travel,
         Tier, N, `% of region`, `Projected N`, `Projected %`)

kable(projection_tbl,
      caption = paste0("Tiered randomizer projection. Tier cutoffs: >=",
                       TIER_CUTOFFS[1], "% / ",
                       TIER_CUTOFFS[2], "-", TIER_CUTOFFS[1], "% / <",
                       TIER_CUTOFFS[2], "%. Combined tier weights (via Qualtrics 5/3/2 slot duplication): ",
                       paste(TIER_WEIGHTS * 100, collapse = "/"), "%."),
      align = c("l", "l", "c", "r", "r", "r", "r")) %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped", "hover")) %>%
  collapse_rows(columns = 1, valign = "top") %>%
  row_spec(which(projection_tbl$`Travel state` == "TOTAL"), bold = TRUE)