r中的某个变量的日期计数

Question

我拥有的：

df1

app	proc proc	rej	old
2021-05-11	2021-05-11	NA	NA
2021-05-11	2021-05-11	NA NA	NA NA
2021-05-11	NA NA	NA	NA
2021-- 05-11	2021-05-11	NA	NA
2021-05-12	2021-05-12	NA	2021-09-12
2021-05-12	2021-05-12	2021-10 2021-10-01	NA
2021-05-13	2021-05 -13	NA	2021-09-12
2021-05-13	2021-05-13	2021-10-01	NA
2021-05-13	2021-05-13	2021-10-10-01	NA
2021-05-13	NA NA	NA NA	NA NA
2021-- 05-13	2021-05-13	NA	2021-09-12
2021-05-13	NA NA	NA	NA NA
2021-05-13	2021-05-13	NA NA NA	NA

我想要什么：一张新桌子，带有App Count of App和New Date。忽略旧的和rej

dfdesired：

日期	应用程序	PROC
2021-05-11	4	3
2021-05-12	2	2
2021-05-13	7	5

Answer 1

尝试此

library(dplyr , warn.conflicts = F)

DF1 |> group_by(App) |> summarise(app = n()) -> df2
DF1 |> group_by(Proc) |> summarise(proc =n()) -> df3

df2  |>  left_join(df3 , by = c("App" = "Proc")) -> DFdesired

输出

DFdesired

# A tibble: 3 × 3
  App          app  proc
  <chr>      <int> <int>
1 2021-05-11     4     3
2 2021-05-12     2     2
3 2021-05-13     7     5

Answer 2

这是另一种使用pivot_longer和pivot_wider的方法，来自tidyr。您可以选择App和proc用于旋转的列，然后删除na的行是值。

library(tidyverse)

df1 %>%
  pivot_longer(cols = c(App, Proc), values_drop_na = T) %>%
  count(value, name) %>%
  pivot_wider(id_cols = value, names_from = name, values_from = n)

输出

  value        App  Proc
  <chr>      <int> <int>
1 2021-05-11     4     3
2 2021-05-12     2     2
3 2021-05-13     7     5