Event table from posting intervals

Turns a table of posting intervals (one row per posting, or per group of postings with the same interval) into the table of dated events consumed by stock_from_events(): each interval adds n to the inflow of its start date and n to the outflow of the day after its end date.

Usage

events_from_intervals(dt, by = NULL, start_col, end_col, n_col = NULL)

Arguments

dt

A data.frame or data.table of intervals.

by

Character vector of grouping columns, or NULL for no grouping.

start_col

Name of the start-date column (class Date or IDate).

end_col

Name of the end-date column (class Date or IDate, inclusive).

n_col

Name of a column of non-negative weights (number of postings per row), or NULL to count each row once.

Value

A data.table with columns by, date (IDate), inflow and outflow, one row per by x date with at least one event, ordered by by and date. Counts are integer when n_col is NULL or integer.

Details

A posting is open on day d when start <= d <= end (end inclusive), so it leaves the stock on end + 1. Events are aggregated by by x date. This helper is meant for tests and small data; on large data the same event table is better computed in the database.

Examples

iv <- data.table::data.table(
  source = c("a", "a", "b"),
  start = data.table::as.IDate(c("2024-01-01", "2024-01-03", "2024-01-02")),
  end = data.table::as.IDate(c("2024-01-02", "2024-01-05", "2024-01-02"))
)
events_from_intervals(iv, by = "source", start_col = "start", end_col = "end")
#>    source       date inflow outflow
#>    <char>     <IDat>  <int>   <int>
#> 1:      a 2024-01-01      1       0
#> 2:      a 2024-01-03      1       1
#> 3:      a 2024-01-06      0       1
#> 4:      b 2024-01-02      1       0
#> 5:      b 2024-01-03      0       1