Skip to contents

Multi-order generative model (MOGen) of sequences

Source: R/memory_wrappers.R
mogen.Rd

Fits Markov chains of every order from 0 to max_order as layers of one multi-order model (higher-order De Bruijn graphs) and selects the order that best balances fit and parsimony.

Usage

mogen(
  data,
  max_order = 5L,
  criterion = c("aic", "bic", "lrt"),
  lrt_alpha = 0.01,
  action = NULL,
  actor = NULL,
  time = NULL,
  session = NULL,
  time_threshold = 900,
  timezone = "UTC"
)

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 Markov order tested. Default 5; capped at the longest sequence minus one, with a message.

criterion

"aic" (default), "bic" or "lrt" (likelihood-ratio test).

lrt_alpha

Significance level of the likelihood-ratio test. Default 0.01.

action, actor, time, session, time_threshold, timezone

Long-format arguments (see sequence-input); leave the column names NULL for wide, list or model input.

Value

A net_mogen object (also a cograph_network). Read it with hg_get(): the order-selection table (what = "orders", the default), the fitted transitions of a layer ("transitions"), the path counts ("paths", with k =) and the pathways ("pathways").

References

Scholtes, I. (2017). When is a network a network? Multi-order graphical model selection in pathways and temporal networks. Proceedings of KDD 2017, 1037–1046. doi:10.1145/3097983.3098145

Gote, C., Casiraghi, G., Schweitzer, F., & Scholtes, I. (2023). Predicting variable-length paths in networked systems using multi-order generative models. Applied Network Science, 8, 68. doi:10.1007/s41109-023-00596-0

See also

markov_order(), memory()

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"))
mg <- mogen(seqs, max_order = 2)
hg_get(mg)
#>   order log_likelihood      aic      bic df layer_df optimal
#> 1     0     -37.457050 82.91410 87.62631  4        4   FALSE
#> 2     1     -12.712215 35.42443 41.31470  5        1   FALSE
#> 3     2      -7.167038 24.33408 30.22434  5        0    TRUE
hg_get(mg, what = "paths", k = 3)
#>   k        path count proportion
#> 1 3 a -> b -> c     4      0.250
#> 2 3 x -> b -> d     4      0.250
#> 3 3 b -> c -> a     2      0.125
#> 4 3 b -> d -> x     2      0.125
#> 5 3 c -> a -> b     2      0.125
#> 6 3 d -> x -> b     2      0.125