Dynamic Feature Generator — dfg

This function generates a package of multiple tables containing dynamic features extracted from a given eventlog.

dfg_pack(
  el,
  period = c("day", "week", "month", "quarter", "year", "sec", "min", "hour"),
  sequential = F,
  event_funs = c("count", "elapsed", "tte", "censored"),
  attr_funs = NULL,
  var_funs = NULL,
  horizon = NULL
)

Arguments

el: (data.frame) The input event-log table. The event-log must have these columns. Column names must match exactly: - caseID, eventType, eventTime, attribute, value evenTime must be of class Date, otherwise, modify accordingly.
period: (character) Choose period of time for which is periodic transition system is required.
sequential: (logical) Should the dynamic features be generated for a sequence of periodic times. If set to FALSE, the dynamic features are generated only for the times at which an event has happend.
event_funs: (character): which types of features do you want to generate from the occurrence of events?