Rank a list of numbers in reverse order with the highest # being ranked as 1-CodePudding

I have the below df

df <-dput(l)
structure(list(l = c(7.4, 7, 6.7, 8.8, 7.3, 9.1, 7.7, 9.5, 7.5, 
7.4, 6.7, 7.3, 7.5)), row.names = c("rankings_message_1", "rankings_message_2", 
"rankings_message_3", "rankings_message_4", "rankings_message_5", 
"rankings_message_6", "rankings_message_7", "rankings_message_8", 
"rankings_message_9", "rankings_message_10", "rankings_message_11", 
"rankings_message_12", "rankings_message_13"), class = "data.frame")

but I would like to add a column to rank the numbers (they are %) where the highest # would be labeled as rank 1 and so on.

I tried this but I get an error:

df$Rank <- rank(desc(df))

CodePudding user response：

You have probably loaded tidyverse without including that in your code. Another way to get there would be

df$Rank <- (nrow(df)   1) - rank(df$l)
df[order(Rank), ]
#                       l Rank
# rankings_message_8  9.5  1.0
# rankings_message_6  9.1  2.0
# rankings_message_4  8.8  3.0
# rankings_message_7  7.7  4.0
# rankings_message_9  7.5  5.5
# rankings_message_13 7.5  5.5
# rankings_message_1  7.4  7.5
# rankings_message_10 7.4  7.5
# rankings_message_5  7.3  9.5
# rankings_message_12 7.3  9.5
# rankings_message_2  7.0 11.0
# rankings_message_3  6.7 12.5
# rankings_message_11 6.7 12.5

CodePudding user response：

I think this would just be

rev(rank(df$l))   # which is wrong, should have been `rank(-df$l)`

The handling of ties is "average" by default but if you wnat ties to have at least one 1 then you have choices. See ?rank

 rev(rank(df$l))
 #[1]  8.5  4.5  1.5  6.5  8.5 13.0 10.0 12.0  4.5 11.0  1.5  3.0  6.5
 ?desc     # you appear to be using a package that is not base R.
#No documentation for ‘desc’ in specified packages and libraries:
#you could try ‘??desc’
 ?rank
 rev(rank(df$l, ties="first"))
 #[1]  9  5  2  7  8 13 10 12  4 11  1  3  6

I couldn't figure out why the 1 came after the 2 and then I started wondering about the correctness of this code. I think it should instead be:

  (df$r1 <- abs( rank(df$l, ties="first") - (length(df$l)  1)) )
  #  [1]  8 11 13  3 10  2  4  1  6  7 12  9  5

Could also rank the negative of the vector:

 df$r2 <- rank(-df$l, ties="last")