R: Create a list of combinations of row names and column names and rank them

Question

I have two matrices:

matrix1:

     col1   col2
row1  5      4

row2  4      6




matrix2:

     col1   col2
row1  48     50

row2  47     46

What I want is a new matrix or table:

          Dim1   Dim2   rank
row2col1   4      47     1
row1col2   4      50     2
row1col1   5      48     3
row2col2   6      46     4

As you can see, I want to rank different combinations of rows and columns first based on dim1, and if there is a tie, by using dim2. The result matrix must be sorted using this rank. How can I achieve it? It is worth noting that matrix1 and 2 contains values for the dim1 and dim2 and have exactly the same column and row names.

Please create a [reproducible example](http://stackoverflow.com/a/5963610/1412059). — Roland, Nov 14 '13 at 14:47

Roland · Accepted Answer · 2013-11-14T15:18:56.403

4

Assuming there are no duplicate rows:

a <- array(c(5,4,4,6,48,47,50,46),c(2,2,2))
dimnames(a) <- list(c("row1", "row2"), c("col1", "col2"), c("m1", "m2"))
# , , m1
# 
# col1 col2
# row1    5    4
# row2    4    6
# 
# , , m2
# 
# col1 col2
# row1   48   50
# row2   47   46

library(reshape2)
b <- acast(melt(a), Var1+Var2~Var3)
b <- b[order(b[,1], b[,2]),]

Or for an arbitrary number of columns:

b <- b[do.call(order, lapply(seq_len(ncol(b)), function(i) b[,i])),]
#add ranks
b <- cbind(b, rank=seq_len(nrow(b)))

#           m1 m2 rank
# row2_col1  4 47    1
# row1_col2  4 50    2
# row1_col1  5 48    3
# row2_col2  6 46    4

edited Nov 14 '13 at 15:18

answered Nov 14 '13 at 15:10

Roland

127,288
10
191
288

how can I do it dynamically using the rownames and column names? – Bob Nov 14 '13 at 15:13
given that the matrix has n rows and m columns. Your example has hard-coded values. Sorry I am a beginner in R :-) – Bob Nov 14 '13 at 15:15
can you also show me how you construct a out of the two matrices dynamically? Thanks! – Bob Nov 14 '13 at 15:23
Look at package `abind`. – Roland Nov 14 '13 at 15:23

score 2 · Answer 2 · answered Nov 14 '13 at 15:10

Have a look at ?order:

df <- data.frame(Dim1=c(6,5,4,4), Dim2=c(46,48,47,50))

o <- order(df$Dim1, df$Dim2)

df[o, ]
#  Dim1 Dim2
#3    4   47
#4    4   50
#2    5   48
#1    6   46

From ?order:

‘order’ returns a permutation which rearranges its first argument into ascending or descending order, breaking ties by further arguments. ... In the case of ties in the first vector, values in the second are used to break the ties. If the values are still tied, values in the later arguments are used to break the tie

R: Create a list of combinations of row names and column names and rank them

2 Answers2