How to group_by() by two variables in a dplyr call?

I am getting the mean by week (wk), however, wk 21 doesn't have data all so I would like to get the mean by 'month' instead. How do I switch group_by(wk) to group_by(month) in the same dplyr call? The final result for wk 21 should be 5.8.

library(dplyr)
fish <- structure(list(wk = c(20, 20, 20, 20, 20, 20, 20, 21, 21, 21, 
21, 21, 21, 21, 22, 22, 22, 22, 22, 22, 22), month = c(5, 5, 
5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 6, 6), pd = c(6, 
4, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 10, 
4, 5, NA, 6)), row.names = c(NA, -21L), class = "data.frame")

fish %>% group_by(wk) %>% summarise(Mean = mean(pd, na.rm=T))

# A tibble: 3 x 2
     wk   Mean
  <dbl>  <dbl>
1    20   5   
2    21 NaN   
3    22   6.25

Solution

You can't have 2 different groupings active, but you could do something like this:

fish |> 
  group_by(month) |>
  mutate(month_mean = mean(pd, na.rm = TRUE)) |>
  group_by(wk) |>
  summarise(Mean = coalesce(mean(pd, na.rm=T), first(month_mean)))
# # A tibble: 3 × 2
#      wk  Mean
#   <dbl> <dbl>
# 1    20  5   
# 2    21  5.8
# 3    22  6.25