---
title: "Lesson 8"
format: html
---


```{r}
require(tidyverse)
library(nycflights13)

flights |>
  select(3:6) |>
  head(3) |>
  rename_all(~ gsub("_", "", .))


flights |>
  select(3:6) |>
  head(3) |>
  rename_with(toupper, .cols = 2:4)

# 每月10号-15号，dep_delay > 100 的航班

# 每月哪个出发地origin的 dep_delay总时长最长

flights |>
  group_by(month, origin) |>
  summarize(n = n(), total_dep_delay = mean(dep_delay, na.rm = TRUE)) |>
  slice_max(total_dep_delay)

# arrange(month, origin, desc(total_dep_delay))

# 每月随机抽取一天，随机抽取三个航班

slice_sample(n = 1)

flights |>
  tidyr::nest(ymddf = -c(year, month, day)) |>
  group_by(year, month) |>
  slice_sample(n = 5) |>
  unnest(ymddf) |>
  group_by(year, month, day) |>
  slice_sample(n = 3)


flights |>
  tidyr::nest(ymddf = -c(year, month, day)) |>
  group_by(year, month) |>
  slice_sample(n = 1) |>
  mutate(
    ymddf = purrr::map(ymddf, \(x) {
      x |>
        slice_sample(n = 3)
    })
  )

# 每月 每个出发地 周末的平均dep_delay 与 工作日的平均dep_delay 差值最大的3个航班

flights |>
  mutate(date = ymd(paste(year, month, day))) |>
  mutate(weekday = wday(date)) |>
  mutate(isworkday = if_else(between(weekday, 2, 6), "Yes", "No")) |>
  group_by(year, month, origin, flight, isworkday) |>
  summarize(mean_delay = mean(dep_delay, na.rm = TRUE)) |>
  tidyr::nest(diffdelaydf = c(isworkday, mean_delay)) |>
  filter(
    purrr::map(diffdelaydf, \(x) {
      nrow(x)
    }) >
      1
  ) |>
  mutate(
    diffdelay = purrr::map_dbl(diffdelaydf, \(x) {
      x |>
        arrange(isworkday) |>
        pull(mean_delay) |>
        diff()
    })
  ) |>
  group_by(year, month, origin) |>
  slice_max(diffdelay, n = 3)


wday(today())

weekday()


slice_sample(n = 1)


flights |>
  group_by(month) |>
  slice_sample(n = 1)
```



```{r}
p1 <- flights |>
  group_by(year, month) |>
  summarize(n = n()) |>
  ggplot(aes(month, n)) +
  geom_point(shape = 21, size = 6, color = "black", fill = "red") +
  geom_line()

ggsave("./a.pdf")
ggsave("./a.png")

require(patchwork)

p1 / p1
```