7

I have this data. I want to calculate Adjacency matrices in R.

How can I do this? V1,V2,V3 are columns.V1 and V2 are NODES, and W3 are weight from V1 to V2. Direction in this data is important. After calculating the Adjacency matrices, I want to calculate shortest path between these vertices with R language.

How can I do this?

      V1      V2     V3
[1] 164885   431072   3
[2] 164885   164885   24
[3] 431072   431072   5
johnsyweb
  • 136,902
  • 23
  • 188
  • 247
sahar
  • 145
  • 2
  • 3
  • 8
  • 2
    What have you tried so far? I see you're new here so perhaps you're not yet aware that Stack Overflow is a place to get answers to programming questions when you try but get **stuck**. At this point it's not clear that you have attempted to tackle the problem. I think the [Stack Overflow FAQ](http://stackoverflow.com/faq) is a good place to start. For questions about R, [this post](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example) is a must-read. – SlowLearner Feb 13 '13 at 09:50

2 Answers2

10

Here is a simpler solution that does not need reshape(). We just create an igraph graph directly from the data frame you have. If you really need the adjacency matrix, you can still get it via get.adjacency():

library(igraph)

## load data
df <- read.table(header=T, stringsAsFactors=F, text=
                 "     V1      V2    V3
                   164885   431072    3
                   164885   164885   24
                   431072   431072    5")

## create graph
colnames(df) <- c("from", "to", "weight")
g <- graph.data.frame(df)
g
# IGRAPH DNW- 2 3 -- 
# + attr: name (v/c), weight (e/n)

## get shortest path lengths
shortest.paths(g, mode="out")
#        164885 431072
# 164885      0      3
# 431072    Inf      0

## get the actual shortest path
get.shortest.paths(g, from="164885", to="431072")
# [[1]]
# [1] 1 2
Gabor Csardi
  • 10,705
  • 1
  • 36
  • 53
9

This should at the least get your started. The simplest way I could think of to get the adjacency matrix is to reshape this and then build a graph using igraph as follows:

# load data
df <- read.table(header=T, stringsAsFactors=F, text="     V1      V2     V3
 164885   431072   3
 164885   164885   24
 431072   431072   5")
> df
#       V1     V2 V3
# 1 164885 431072  3
# 2 164885 164885 24
# 3 431072 431072  5

# using reshape2's dcast to reshape the matrix and set row.names accordingly
require(reshape2)
m <- as.matrix(dcast(df, V1 ~ V2, value.var = "V3", fill=0))[,2:3]
row.names(m) <- colnames(m)

> m
#        164885 431072
# 164885     24      3
# 431072      0      5

# load igraph and construct graph
require(igraph)
g <- graph.adjacency(m, mode="directed", weighted=TRUE, diag=TRUE)
> E(g)$weight # simple check
# [1] 24  3  5

# get adjacency
get.adjacency(g)

# 2 x 2 sparse Matrix of class "dgCMatrix"
#        164885 431072
# 164885      1      1
# 431072      .      1

# get shortest paths from a vertex to all other vertices
shortest.paths(g, mode="out") # check out mode = "all" and "in"
#        164885 431072
# 164885      0      3
# 431072    Inf      0
Arun
  • 116,683
  • 26
  • 284
  • 387