The session behind each sequence

Names every sequence of a network or of a clustering by the columns it was built from. A network built from long data with actor and session has one sequence per actor-session; its rows are ordered by the grouping, not by the input, and a fitted clustering reports its assignments in that same order. session_ids() returns the key that joins them back to the input data, so no label has to be parsed.

Usage

session_ids(x, ...)

# Default S3 method
session_ids(x, ...)

# S3 method for class 'netobject'
session_ids(x, ...)

# S3 method for class 'net_mmm'
session_ids(x, ...)

# S3 method for class 'net_clustering'
session_ids(x, ...)

Arguments

x

A netobject from build_network on long data, or a net_mmm (build_mmm) or net_clustering (build_clusters) fitted on such a network.

...

Unused.

Value

A data frame with one row per sequence, in the row order of the network's $data (and of the fit's assignments):

sequence

Integer row number of the sequence.

actor and session columns

The actor and session columns given to build_network(), under their own names and with their own values.

session_label

The readable label of the sequence. With time, sessions split at time gaps carry a " s<n>" suffix, so this column separates them.

cluster

For a net_mmm or net_clustering: the assigned cluster (integer).

posterior

For a net_mmm: the posterior probability of the assigned cluster.

Errors

Raises nestimate_no_session_ids when x carries no per-sequence metadata: a network built from wide data, a fit on wide data or on a tna model, or a fit made before Nestimate 0.9.6 (refit it). Raises nestimate_session_ids_misaligned when the metadata and the sequences differ in number.

See also

build_network, build_mmm, build_clusters

Examples

events <- data.frame(
  student = rep(c("s1", "s2", "s3"), each = 8),
  step    = rep(c("a", "b"), each = 4, times = 3),
  action  = sample(c("read", "write", "test"), 24, replace = TRUE)
)
net <- build_network(events, actor = "student", session = "step",
                     action = "action", method = "relative")
session_ids(net)
#>   sequence student step session_label
#> 1        1      s1    a        s1 | a
#> 2        2      s2    a        s2 | a
#> 3        3      s3    a        s3 | a
#> 4        4      s1    b        s1 | b
#> 5        5      s2    b        s2 | b
#> 6        6      s3    b        s3 | b

# \donttest{
fit <- build_mmm(net, k = 2, n_starts = 2, seed = 1)
session_ids(fit)
#>   sequence student step session_label cluster posterior
#> 1        1      s1    a        s1 | a       2 0.9999906
#> 2        2      s2    a        s2 | a       1 0.9999245
#> 3        3      s3    a        s3 | a       1 0.9999830
#> 4        4      s1    b        s1 | b       2 0.8881216
#> 5        5      s2    b        s2 | b       1 1.0000000
#> 6        6      s3    b        s3 | b       1 0.9999990
# }