Transform processed relational event sequences to different formats

This vignette explains input arguments, output structure and usage of the function remify::rehshape().

---

Aim

remify::rehshape() transforms a remify object into another object with a structure that is suitable to external packages. The function can return the data inputs required by the functions:

• relevent::rem()
• relevent::rem.dyad()

Both functions are available inside the relevent package (Butts C., 2023).

---

Input

The input arguments of remify::rehshape() are:

• data, the processed relational event history (S3 object of class remify)
• output_format, a character value that indicates to which output format the input data should be converted. This argument can assume two values: "relevent-rem" , "relevent-rem.dyad" (default is "relevent-rem")
• ncores, number of threads used to parallelize internal routines (default is 1L)
• optional_arguments, vector of arguments names from relevent::rem or relevent::rem.dyad() that the user might want to process and have in the output object of rehshape (e.g., the pre-computed structures required by relevent::rem.dyad(), such as acl, cumideg, etc.) - this feature will be available in a future version of remify -

---

Output

The output structure of the function is different according to the chosen output_format:

• If output_format = "relevent-rem", then the output is an S3 object of class relevent-rem, which contains:
  • eventlist, a matrix of two columns: observed dyads in the first column, vector of time in the second column
  • supplist, a logical matrix of dimensions [rows = number of events, columns = number of dyads]. The matrix indicates at each time point (by row) whether each dyad was at risk (TRUE), or not (FALSE)
  • timing, is a character that can assume two values: "interval" (which uses the inter-event time in the model), or "ordinal" (which only considers the event order in the model)
• If output_format = "relevent-rem.dyad", then the output is an S3 object of class relevent-rem.dyad, which contains:
  • edgelist, a matrix of three columns: the time (or order) of the events in the first column, the sender and the receiver of the relational event, respectively, in the second and third column
  • n, is the number of actors in the relational event network (senders and receivers)
  • ordinal, is a logical (TRUE/FALSE) value which indicates whether the likelihood should be ‘ordinal’ (TRUE) or ‘interval’ (FALSE)

---

Usage

To explain the usage of the function remify::rehshape(), we consider the example edgelist available with the data randomREH. First, we process the edgelist with remify::remify().

library(remify)
data(randomREH) 
reh_remify <- remify::remify(edgelist = randomREH$edgelist, model = "tie")
reh_remify

## Relational Event Network
## (processed for tie-oriented modeling):
##  > events = 9915
##  > actors = 20
##  > (event) types = 3
##  > riskset = full
##  > directed = TRUE
##  > ordinal = FALSE
##  > weighted = FALSE
##  > time length ~ 80 days
##  > interevent time 
##       >> minimum ~ 0.0011 seconds
##       >> maximum ~ 5811.4011 seconds

Then, we can transform the remify object to any of the possible output formats:

• from remify to relevent-rem:

reh_rem <- remify::rehshape(data = reh_remify, 
                      output_format = c("relevent-rem"))
names(reh_rem)                     

## [1] "eventlist" "supplist"  "timing"

• from remify to relevent-rem.dyad:

reh_rem.dyad <- remify::rehshape(data = reh_remify, 
                      output_format = c("relevent-rem.dyad"))
names(reh_rem.dyad)                     

## [1] "edgelist" "n"        "ordinal"

---