Detect career transitions

Collapses each entity's history into spells of consecutive periods in the same group, then pairs each spell with the one that follows it. A row is dated by ref_date, the period in which the entity is first observed in the destination group, so a transition is counted when it happens. from_date records when the origin spell began.

Usage

detect_career_transition(
  data,
  id_col = "contract_id",
  group_cols,
  return_all = FALSE
)

Arguments

data

Data frame containing a ref_date column, the identifier and the grouping columns.

id_col

Character. Column identifying the entity whose career is tracked. Default "contract_id".

group_cols

Character vector of columns defining the career position (e.g. paygrade, department). Multiple columns are pasted into one label.

return_all

Logical. Keep terminal spells, which have no destination and so carry NA in both to and ref_date. Default FALSE.

Value

A data table with the identifier, from, to, from_date and ref_date.

Details

Records must be uniquely identified by id_col and ref_date. An entity holding more than one record in the same period has no well-defined position, so every record for that entity is dropped with a warning reporting how many.