在这里，我重塑了long，将雇用计数为+1，将终止计数为-1，按日期汇总，然后将计数计算为雇用和终止的累积总和。（一个可能的变化可能是将终止日期调整为晚一天-例如，如果有人工作了一天，我们可能会认为他们早上是+1，下午是-1，所以他们当天贡献了+1，而不是零。我在这里没有做这个调整）。

library(tidyverse)
df1 |>
  pivot_longer(1:2) %>%
  mutate(change = if_else(name == "original_hire_date", 1, -1),
         date = lubridate::mdy(value)) %>%
  filter(!is.na(date)) |>
  arrange(date) |>
  count(date, wt = change, name = "change") |>
  complete(date = seq.Date(min(date), max(date), by = 1),  # to fill in all days,
           fill = list(change = 0)) |> # so count doesn't drift between observations
  mutate(count = cumsum(change)) |>
  ggplot(aes(date, count)) +
  geom_line()

字符串
x1c 0d1x的数据

这里有一个方法：

library(tidyverse)

grouped_df <- df %>% 
  # put the hirings and firing dates into one column
  pivot_longer(cols = everything(), names_to = "status", values_to = "date", values_drop_na = TRUE) %>%

  # parse the dates as date format
  mutate(date = ymd(format(mdy(date), "%Y/%m/01"))) %>%
  count(status, date) %>%

  # pivot it wider again, so each year-month combination has it's own column
  pivot_wider(names_from = status, values_from = n, values_fill = 0) %>%

  # get the change and cumulative sum of workers
  mutate(`Net Change` = original_hire_date - termination_date, 
         `Total Workers` = cumsum(`Net Change`))

grouped_df %>%
  ggplot(aes(x = date, y = `Total Workers`)) +
  geom_line() + 
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Total Workers")

字符串

的数据