R simple summary table-CodePudding

level_A <- c(1,2,3,4,5)
level_B <- c(1,2,3,4,5)
year <- c(2001,2002,2003,2001,2002)

df <- as.data.frame(cbind(level_A,level_B, year))

df %>% 
  group_by(level_A, level_B, year) %>% 
  summarise(n=n()) %>%
  spread(year, n)%>%
  adorn_totals()

table(df$level_A, df$year)

I want to make a table

How can I make this with tidyverse?

CodePudding user response：

reshape2

reshape2::dcast(
  reshape2::melt(df, id.vars = "year"),
  variable ~ year, value.var = "value", fun.aggregate = length)
#   variable 2001 2002 2003
# 1  level_A    2    2    1
# 2  level_B    2    2    1

And ... now I've learned a new function in the package (thanks to onyambu), doing the same thing in one step:

reshape2::recast(df, variable~year, id.var = 'year', fun.ag=length)
#   variable 2001 2002 2003
# 1  level_A    2    2    1
# 2  level_B    2    2    1

CodePudding user response：

Here's a possible tidyverse solution, where we first pivot to long form, then summarise the number of observations, then pivot to a wide format.

library(tidyverse)

df %>%
  pivot_longer(starts_with("level"), names_to = "Var") %>%
  group_by(year, Var) %>%
  summarise(value = sum(!is.na(value))) %>%
  pivot_wider(names_from = year, values_from = value)

Output

  Var     `2001` `2002` `2003`
  <chr>    <int>  <int>  <int>
1 level_A      2      2      1
2 level_B      2      2      1