path_counts() counts how often each path of k consecutive states
(a k-gram) occurs across all trajectories, and returns one row per path,
most frequent first. Every position of a trajectory starts a path, so the
paths of one trajectory overlap. The counts are descriptive: they show
which runs of states dominate the data before a model such as
build_mogen() or build_hon() is fitted to them.
Arguments
- data
The trajectories: a list of state vectors, a data frame or character matrix with one row per trajectory and one column per time point, a
tnamodel or anetobjectbuilt from sequences. A data frame needs position columns named as a numbered run (T1,T2, ...); any other frame, a one-hot or association network and a 0/1 matrix raise an error of classnestimate_bad_input. A missing value ends a run of observed states, and no path spans it.- k
The number of states in each path, a whole number of at least 2; a non-integer value is an error, not truncated. Default
2, which counts single transitions;3counts two-step paths, and so on.- top
The number of most frequent paths to keep. Default
NULL, every path.
Value
A data frame with one row per distinct path, sorted by count
(descending), with columns path (the k states in arrow notation, such
as "A -> B"), count and proportion (the path's share of all paths
counted, rounded to 4 decimal places). When no trajectory holds k
consecutive observed states, the data frame has no rows.
Examples
trajs <- list(c("A","B","C","D"), c("A","B","D","C"))
path_counts(trajs, k = 2)
#> path count proportion
#> 1 A -> B 2 0.3333
#> 2 B -> C 1 0.1667
#> 3 B -> D 1 0.1667
#> 4 C -> D 1 0.1667
#> 5 D -> C 1 0.1667
# \donttest{
path_counts(trajs, k = 3, top = 10)
#> path count proportion
#> 1 A -> B -> C 1 0.25
#> 2 A -> B -> D 1 0.25
#> 3 B -> C -> D 1 0.25
#> 4 B -> D -> C 1 0.25
# }