Sequence pattern matching

Counts, per sequence, how many times each of up to 20 patterns occurs – TDA's seqpm. A pattern is a vector of states and wildcards, one element per position: c(3, 3) looks for state 3 immediately followed by state 3; "?" matches any one state, "*" any run of states (including none), "+" any repeat of the state before it, and "-" any run of identical states. Several patterns are given as a list, list(c(3, 3), c(3, "*", 3)).

Usage

tda_seqpm(
  sequences,
  patterns,
  variables = NULL,
  id = NULL,
  options = list(),
  dir = tempfile("tda"),
  ...
)

Arguments

sequences

a data frame or matrix of one column per time point, in order – the same shape tda_seqm takes – optionally with further, non-sequence columns (an id, say) alongside them; use variables/id to say which is which.

patterns

a list of patterns, each a vector of states (numbers) and wildcards ("?", "*", "+", "-", as strings), as described above. At most 20.

variables

which columns of sequences are the actual time-point columns, in order – TDA's seqdef=Y0,,Y7, which likewise names the sequence out of a data matrix holding other variables too. A character vector of names in sequences; defaults to every column (so an already-subsetted sequences still works unchanged).

id

further columns of sequences to carry through to the output as-is, one value per case – seqpm's v=. A character vector of names in sequences.

options

a named list of further TDA options, passed through.

dir

working directory.

...

passed to tda_run.

Value

An object carrying a table with one row per case: its sequence id, its length, one match-count column per pattern, and any id columns requested.

See also

Other sequence analysis: tda_evalfi(), tda_fml(), tda_frml(), tda_seq_info(), tda_seqgc(), tda_seqm(), tda_seqmd(), tda_seqpe()

Examples

# six short sequences over three states; does "3,3" (state 3 twice in a
# row) occur, and does the sequence ever return to 1 after leaving it?
s <- data.frame(Y1 = c(1, 1, 3, 3, 3, 1), Y2 = c(1, 3, 3, 3, 1, 2),
                Y3 = c(1, 3, 3, 1, 2, 2), Y4 = c(2, 3, 1, 2, 2, 2),
                Y5 = c(2, 1, 2, 2, 2, 3))
r <- tda_seqpm(s, list(c(3, 3), c(1, "*", 1)))
r$table
#>   case length matches1 matches2
#> 1    1      5        0        1
#> 2    2      5        1        1
#> 3    3      5        1        0
#> 4    4      5        1        0
#> 5    5      5        0        0
#> 6    6      5        0        0

# an id column alongside the sequence, positively selected
s2 <- cbind(person = 101:106, s)
r2 <- tda_seqpm(s2, list(c(3, 3)), variables = names(s), id = "person")
r2$table
#>   case length matches1 Vperson
#> 1    1      5        0     101
#> 2    2      5        1     102
#> 3    3      5        1     103
#> 4    4      5        1     104
#> 5    5      5        0     105
#> 6    6      5        0     106