Higher-order network (HON) from sequences

Builds a higher-order network: a state is split into memory nodes ("a -> b": state b reached from a) wherever the next-state distribution depends on where the sequence came from, as decided by the BuildHON / BuildHON+ rule extraction.

Usage

hon(
  data,
  max_order = 5L,
  min_freq = 1L,
  collapse_repeats = FALSE,
  method = "hon+",
  action = NULL,
  actor = NULL,
  time = NULL,
  session = NULL,
  time_threshold = 900,
  timezone = "UTC",
  group = NULL
)

Arguments

data: Sequences in any form described in sequence-input: a long event table (with action), a wide data.frame, a list of vectors, or a model object carrying its sequences.
max_order: Integer. Highest order considered. Default 5.
min_freq: Integer. Transitions observed fewer times are treated as zero. Default 1.
collapse_repeats: Logical. Collapse adjacent repeated states before extraction. Default FALSE.
method: "hon+" (default, parameter-free BuildHON+) or "hon" (the original eager BuildHON).
action, actor, time, session, time_threshold, timezone: Long-format arguments (see sequence-input); leave the column names NULL for wide, list or model input.
group: NULL (default) for one network, or the grouping of the sequences for a group model: the name of a column of data (long or wide format), or a vector with one label per sequence (per row of a wide data frame, per element of a list). A group model holds one network per group and the data of each group; it is the input of hg_compare() and can be passed to hg_bootstrap() and hg_get().

Value

A net_hon object (also a cograph_network, so it plots directly). Read its tables with hg_get(): the rules (what = "rules", the default), the memory nodes ("nodes") and the higher-order pathways ("pathways").

References

Xu, J., Wickramarathne, T. L., & Chawla, N. V. (2016). Representing higher-order dependencies in networks. Science Advances, 2(5), e1600028. doi:10.1126/sciadv.1600028

Saebi, M., Xu, J., Kaplan, L. M., Ribeiro, B., & Chawla, N. V. (2020). Efficient modeling of higher-order dependencies in networks: from algorithm to application for anomaly detection. EPJ Data Science, 9, 15. doi:10.1140/epjds/s13688-020-00233-y

Examples

seqs <- list(c("a", "b", "c", "a", "b", "c"), c("x", "b", "d", "x", "b", "d"),
             c("a", "b", "c", "a", "b", "c"), c("x", "b", "d", "x", "b", "d"))
net <- hon(seqs, max_order = 2)
hg_get(net)
#>          path   from to count probability from_order to_order
#> 1      a -> b      a  b     4         1.0          1        2
#> 2 a -> b -> c a -> b  c     4         1.0          2        1
#> 3      b -> c      b  c     4         0.5          1        1
#> 4      b -> d      b  d     4         0.5          1        1
#> 5      c -> a      c  a     2         1.0          1        1
#> 6      d -> x      d  x     2         1.0          1        1
#> 7      x -> b      x  b     4         1.0          1        2
#> 8 x -> b -> d x -> b  d     4         1.0          2        1

# the same network from a long event table
events <- data.frame(
  who  = rep(c("s1", "s2", "s3", "s4"), each = 6),
  step = rep(1:6, 4),
  what = unlist(seqs)
)
hg_get(hon(events, action = "what", actor = "who", time = "step",
           max_order = 2), what = "nodes")
#> Error in Nestimate::prepare(data[columns], actor = actor, action = action,     time = time, session = session, time_threshold = time_threshold,     timezone = timezone): unused argument (timezone = timezone)