Vstupní analýza struktury dat

dtl %>% 
  ungroup() %>% 
  summarise(length(unique(prj_id)))

dt_geostructure_byprj <- dtl %>% 
  filter(!is.na(value) & typ == "id") %>% 
  # sample_n(10) %>% 
  # mutate(level_num = as.numeric(level)) %>% 
  select(prj_id, level, op_id, value) %>% 
  group_by(prj_id, op_id) %>% 
  summarise(unique_units = length(unique(value)),
            n_levels = length(unique(level)),
            maxlevel = max(level), minlevel = min(level), 
            n_units = n(),
            .groups = "drop") %>% 
  mutate(multilevel = n_levels > 1) %>% 
  ungroup()

write_parquet(dt_geostructure_byprj, here::here("data-processed", "dt_geostructure_by-prj.parquet"))

dt_geostructure_bylvl <- dtl %>% 
  filter(!is.na(value) & typ == "id") %>% 
  select(prj_id, level, op_id, value) %>% 
  group_by(prj_id, op_id, level) %>% 
  summarise(n_units = length(unique(value))) %>%
  group_by(prj_id, op_id) %>%
  mutate(min_level = level == min(level)) %>% 
  mutate(multiunit = n_units > 1) %>%
  ungroup()

write_parquet(dt_geostructure_bylvl, here::here("data-processed", "dt_geostructure_by-lvl.parquet"))

prj_no_obec <- dtl %>%
  filter(typ == "id") %>%
  drop_na(value) %>%
  group_by(prj_id) %>%
  summarise(noobecnozuj = all(!(level %in% c("zuj", "obec"))), .groups = "drop") %>%
  filter(noobecnozuj) %>%
  select(prj_id)

dim(prj_no_obec)

write_parquet(prj_no_obec, here::here("data-processed", "prj_id_noobec.parquet"))

dt_geostructure_byprj %>% 
  mutate(notoplevel = prj_id %in% prj_no_obec$prj_id) %>% 
  count(op_id, wt = mean(notoplevel)) %>%
  bind_rows(dt_geostructure_byprj %>% 
              mutate(notoplevel = prj_id %in% prj_no_obec$prj_id) %>% 
              count(wt = mean(notoplevel)) %>% 
              mutate(op_id = "Celkem")) %>% 
  mutate(op_id = fct_reorder(op_id, n)) %>% 
  ggplot(aes(n, op_id)) +
  geom_col() +
  theme_ptrr("x") +
  scale_x_percent_cz(limits = c(0, 1))

library(ggupset)

dtl_for_upset <- dtl %>% 
  filter(typ == "id" & !is.na(value)) %>% 
  select(op_id, prj_id, level) %>%
  distinct() %>% 
  mutate(level = as.character(level)) %>% 
  group_by(prj_id) %>% 
  summarise(geos = list(level)) %>%
  left_join(dt %>% filter(prj_radek == 1) %>% select(prj_id, op_id)) %>% 
  # mutate(nr = map_int(geos, nrow)) %>%
  # arrange(desc(nr)) %>% 
  filter(TRUE)

ggplot2::update_geom_defaults("line", list(colour = "black"))
ggplot(dtl_for_upset) +
  geom_bar(mapping = aes(x = geos, fill = op_id), position = "stack") +
  scale_x_upset(order_by = "degree", sets = geolevels) +
  # scale_y_continuous(limits = c(0, 6e4)) +
  # scale_y_log10(limits = c(1, 1e5), n.breaks = 10,
  #               labels = scales::label_number(1)) +
  # scale_y_continuous(trans = scales::log_trans(), breaks = scales::pretty_breaks()) +
  ptrr::theme_ptrr() +
  scale_fill_brewer(type = "qual")

ggplot2::update_geom_defaults("line", list(colour = "darkblue"))

ggplot2::update_geom_defaults("line", list(colour = "black"))

ggplot(dtl_for_upset) +
  geom_bar(mapping = aes(x = geos)) +
  scale_x_upset(order_by = "degree", sets = geolevels) +
  # scale_y_log10(n.breaks = 6, labels = scales::label_number(1)) +
  # scale_y_continuous(trans = scales::log_trans(), breaks = scales::pretty_breaks()) +
  ptrr::theme_ptrr(multiplot = T) +
  facet_wrap(~op_id, scales = "free")

ggplot2::update_geom_defaults("line", list(colour = "darkblue"))

dt_geostructure_byprj %>% 
  count(op_id, maxlevel) %>% 
  mutate(maxlevel = fct_relevel(maxlevel, geolevels)) %>% 
  spread(maxlevel, n)

dtl %>%  
  filter(!is.na(value) & typ == "id") %>% 
  distinct(prj_id, level)

multilevel <- dtl %>% 
  left_join(dt_geostructure_byprj %>% select(prj_id, multilevel)) %>%  
  ungroup()

  # filter(multilevel) %>%

multilevel_ids <- unique(multilevel[multilevel$multilevel,]$prj_id)

multilevel %>% 
  distinct(op_id, prj_id, multilevel) %>% 
  summarise(podil = mean(multilevel))

multilevel %>% 
  distinct(op_id, prj_id, multilevel) %>% 
  group_by(op_id) %>% 
  summarise(podil = mean(multilevel), .groups = "drop")