CODEWITHSUNDEEP
rfunctionmatrixtransition
user2646205
user2646205

Reputation: 5

Calculation of square transition matrix

I want to calculate square de-normalized (= frequencies) transition matrix between state p(=previous) and n(=next). I have tried following script:

p <- c("S1", "S1", "S2", "S3", "S4")
n <- c("S2", "S2", "S3", "S4", "S5")
d <- data.frame(fprev=p, fnext=n)
counts <- table( as.data.frame( d ))

counts is following:

     fnext
fprev S2 S3 S4 S5
   S1  2  0  0  0
   S2  0  1  0  0
   S3  0  0  1  0
   S4  0  0  0  1

This is quite close what I am wanting but I would like to get similar structure except that values for both fnext and fprev would be S1 S2 S3 S4 S5 (I have tried dnn parameter to table and deparse.level but they did not seem to help)

So expected output should have S1 S2 S3 S4 S5 in both column and row directions (naturally some more values 0 as for example there is no S5 state in fprev)

Upvotes: 0

Views: 251

Answers (1)

Stuples
Stuples

Reputation: 1446

In order to get zeroes for all levels in your table you should set fprev and fnext to be factors with the same levels like so

p <- c("S1", "S1", "S2", "S3", "S4")
n <- c("S2", "S2", "S3", "S4", "S5")

lvls <- sort(unique(c(p, n)))

d <- data.frame(
    fprev = factor(p, levels=lvls),
    fnext = factor(n, levels=lvls)
)

counts <- table(d)

Now your counts table should be how you want it.

fprev S1 S2 S3 S4 S5
   S1  0  2  0  0  0
   S2  0  0  1  0  0
   S3  0  0  0  1  0
   S4  0  0  0  0  1
   S5  0  0  0  0  0

Upvotes: 3

Related Questions