verifyEmpiricalToTheoretical Errors

[explore-starling]

I've been running into issues using the verifyEmpiricalToTheoretical() function.

library(markovchain)

states <- c("LRP", "LF", "LFP", "Bi", "RFP", "RF", "RRP", "BS", "ES") # list of possible states
transitions <- matrix(c(0.1639, 0.0459, 0.0852, 0.0569, 0.0852, 0.0459, 0.1639, 0.1399, 0.2132,
              0.1639, 0.0459, 0.0852, 0.0569, 0.0852, 0.0459, 0.1639, 0.1399, 0.2132,
              0.1639, 0.0459, 0.0852, 0.0569, 0.0852, 0.0459, 0.1639, 0.1399, 0.2132,
              0.1639, 0.0459, 0.0852, 0.0569, 0.0852, 0.0459, 0.1639, 0.1399, 0.2132,
              0.1639, 0.0459, 0.0852, 0.0569, 0.0852, 0.0459, 0.1639, 0.1399, 0.2132,
              0.1639, 0.0459, 0.0852, 0.0569, 0.0852, 0.0459, 0.1639, 0.1399, 0.2132,
              0.1639, 0.0459, 0.0852, 0.0569, 0.0852, 0.0459, 0.1639, 0.1399, 0.2132,
              0.1639, 0.0459, 0.0852, 0.0569, 0.0852, 0.0459, 0.1639, 0.1399, 0.2132,
              0.2083, 0.0583, 0.1084, 0.0722, 0.1084, 0.0583, 0.2083, 0.1778, 0), byrow=TRUE,
            nrow = 9, ncol=9, dimnames = list(c("LRP", "LF", "LFP", "Bi", "RFP", "RF", "RRP", "BS", "ES"),c("LRP", "LF", "LFP", "Bi", "RFP", "RF", "RRP", "BS", "ES")))
# creates theoretical transition matrix


chainE <- new("markovchain", states = states, transitionMatrix = transitions, 
              name = "ChainE") # creates markov chain object for theoretical transitions

# to create sample empirical sequence
generate_sequence <- function(states, n) {
  result <- character(n)
  prev <- ""

   for (i in 1:n) {
   # If previous state was "ES", exclude it from sampling
   options <- if (prev == "ES") states[states != "ES"] else states

   result[i] <- sample(options, 1)
   prev <- result[i]
 }
 return(result)
}

seq <- generate_sequence(states, 46000) # creates a sample like my dataset

verifyEmpiricalToTheoretical(data = seq, object = chainE, verbose = TRUE) # should compare the sequence and the theoretical markov chain?

I got some help over on StackOverflow (https://stackoverflow.com/questions/79560561/verifyempiricaltotheoretical-returns-error-even-when-using-character-sequence) to fix the first issue, which is that the function was not working despite my seq vector being characters.

verifyEmpiricalToTheoretical2 <- verifyEmpiricalToTheoretical
body(verifyEmpiricalToTheoretical2)[[4]][[2]] <- quote(!(is.numeric(data) || is.character(data) || is.matrix(data)))
environment(verifyEmpiricalToTheoretical2) <- environment(verifyEmpiricalToTheoretical)
verifyEmpiricalToTheoretical2(data = seq, object = chainE, verbose = TRUE)

With that code, I'm able to get an output, but it seems to be incorrect.

"At least one transition is >0 in the data that is 0 in the object. Therefore the null hypothesis is rejected."

However, the only transition in my sequence matrix that is 0 is also 0 in the transition matrix (this is because by definition, ES cannot follow ES in my data).

[spedygiorgio]

dear @explore-starling thanks for the point raisen. I know the issue but unfortunately I lack resources to fix it. Should you perform a pull request to fix it, I would be happy to overview it and eventually merge in the main branch.