Looking for a function to fined the mean of the top n values (not rows!) conditionally and return the number and not a dataframe

I have a large data frame: percentage_activity

# A tibble: 4,437 x 3
# Groups:   DATETIME [87]
   DATETIME            ID        COUNT
   <dttm>              <chr>     <int>
 1 2020-06-07 00:00:00 Bagheera     NA
 2 2020-06-07 00:00:00 Bagheera2     0
 3 2020-06-07 00:00:00 Baloo img     0
 4 2020-06-07 00:00:00 Banna        NA
 5 2020-06-07 00:00:00 Blair       158
 6 2020-06-07 00:00:00 Carol        NA

in which I would like to calculate the mean of the top 5 COUNTs for a specific ID, and then, in a for loop, represent every COUNT value as a quantity with the mean value calculated for this ID as the 100% of this specific ID. To do that, I would really rather get a mean value not as a datafrme for all individuals but as a single number for the desired ID, and then use it as a variable inside the for loop.

I'm actually trying to reconstruct a loop that workd for the same data orgenized with seperated columns for each ID, but after melting the data to one ID colum It needs adjusments:

max_activity <- readline(prompt="enter a number: ")
    for(i in 2:length(percentage_activity)) {
    percentage_activity[[i]] <- 
     as.numeric(percentage_activity[[i]]*100/mean(sort(percentage_activity[[i]] ,T) 
    [1:max_activity]))
}

I also tried this, but I'm not sure how to proceed from here:

for (i in unique(percentage_activity$ID)){
  individual <- percentage_activity$ID == i
  mean(percentage_activity[individual,"COUNT"], na.rm=TRUE)
}

Maybe this may help:

library(dplyr)
df <- tibble(
  DATETIME = as.Date(c("2020-06-07",
                       "2020-06-07",
                       "2020-06-07",
                      "2020-06-07",
                       "2020-06-07",
                       "2020-06-07",
                      "2020-06-07",
                      "2020-06-07",
                      "2020-06-07",
                      "2020-06-07",
                      "2020-06-07",
                      "2020-06-07")),
  ID = c("Bagheera", "Bagheera2", "Baloo img", "Banna", "Blair", "Carol", 
         "Bagheera", "Bagheera2", "Baloo img", "Banna", "Blair", "Carol"),
  COUNT = c(NA, 0,0,NA, 158, NA,10,20,30,40,50, 60)
)

mean_val <- df %>% 
  group_by(ID) %>% 
  arrange(desc(COUNT)) %>% 
  top_n(5) %>% 
  summarise(mean = mean(COUNT, na.rm = T)) 

df %>% 
  left_join(mean_val, by = "ID") %>% 
  mutate(percentage_activity =  COUNT/mean)