Sort a factor based on value in one or more other columns

Here's a reproducible sample, with solution:

set.seed(0)
a = sample(1:20,replace=F)
b = sample(1:20,replace=F)
f = as.factor(letters[1:20])

> a
 [1] 18  6  7 10 15  4 13 14  8 20  1  2  9  5  3 16 12 19 11 17
> b
 [1] 16 18  4 12  3  5  6  1 15 10 19 17  9 11  2  8 20  7 13 14
> f
 [1] a b c d e f g h i j k l m n o p q r s t
Levels: a b c d e f g h i j k l m n o p q r s t

Now for the new factor:

fn = factor(f, levels=unique(f[order(a,b,f)]), ordered=TRUE)

> fn
 [1] a b c d e f g h i j k l m n o p q r s t
20 Levels: k < l < o < f < n < b < c < i < m < d < s < q < g < h < e < ... < j

Sorted on 'a', next 'b' and finally 'f' itself (although in this example, 'a' has no repeated values).

I recommend the following dplyr-based approach (h/t daattali) that can be extended to as many columns as you like:

library(dplyr)
Catalog <- Catalog %>%
  arrange(MIDDATE, TYPENAME) %>%               # sort your dataframe
  mutate(IDENTIFY = factor(IDENTIFY, unique(IDENTIFY))) # reset your factor-column based on that order