library(tidyverse)
library(babynames)

# 1)
babynames %>%
    filter(year >= 2000) %>%
    group_by(sex, name) %>%
    summarize(n=sum(n), .groups="drop") ->
    bn20

# 2)
bn20 %>%
    group_by(name) %>%
    pivot_wider(names_from="sex", values_from="n") %>%
    rename(female = "F", male = "M") %>%
    filter(female > 0 & male > 0 & (female+male) > 1e5) %>%
    arrange(abs(female - male) / (male + female)) ->
    bn20_sim

# 3)
bn20 %>% group_by(name_length = nchar(name)) %>% summarize(n=sum(n))

# 4)  
min_len <- min(nchar(bn20$name))
bn20 %>% filter(nchar(name) == min_len) %>% arrange(desc(n))

# 5)
babynames %>% 
   group_by(year, sex) %>%
   summarize(avg_len = mean(nchar(name))) ->
   bn_avg_len

bn_avg_len %>%
    ggplot() +
    geom_line(aes(x=year, y=avg_len, color=sex))
  
# 6)
babynames %>%
    select(-prop) %>%
    filter(year >= 2000) ->
    bn

# Annahme: "Neue" Namen die im Vorjahr noch nicht in bn auftauchen
# werden ausgewertet mit n_prev = 0
replace_with_zero <- function(x) {
    x[is.na(x)] <- 0
    x
}

bn %>%
    filter(year > 2000) %>%
    left_join(mutate(bn, year = year + 1, n_prev = n, n = NULL)) %>%
    mutate(n_prev = replace_with_zero(n_prev)) ->
    bn_prev

# Wenn das nicht erwünscht, führe man diese Zeile aus
# bn %>% inner_join(mutate(bn, year = year + 1, n_prev = n, n = NULL)) -> bn_prev

n_mean <- mean(bn$n)

bn_prev %>% 
    mutate(s_incr = (n-n_prev)/(n_prev + n_mean)) %>%
    group_by(year, sex) %>%
    slice_max(s_incr, n = 3) ->
    bn_trending