Skip to contents

Derive monthly / quarterly / semi-annual / annual grain columns

Source: R/period.R
derive_grain_columns.Rd

Given a long-format frame with monthly source columns (uy_m, cy_m, optionally dev_m), derive the coarser-grain siblings (uy_q / uy_h / uy, cy_q / cy_h / cy, dev_q / dev_h / dev_y) so the same frame can be aggregated at any of the four grains.

This is an optional utility – as_triangle() and as_calendar() already derive the single grain they need internally. Use this when you want a single enriched frame that can be re-aggregated at multiple grains, or for exploratory plots.

Usage

derive_grain_columns(df)

Arguments

df

A data.frame containing uy_m, cy_m, and optionally dev_m. Coarser-grain siblings are derived from these.

Value

A data.table with the additional grain columns.

Details

Letter-suffix family: _m / _q / _h / _y = monthly / quarterly / half-yearly / yearly.

Derived columns when source columns exist:

Underwriting (from uy_m):

  • uy : yearly start (Jan 1 of uy_m's year)

  • uy_h : half-yearly start (Jan 1 / Jul 1)

  • uy_q : quarterly start (Jan / Apr / Jul / Oct 1)

Calendar (from cy_m):

  • cy : yearly start

  • cy_h : half-yearly start

  • cy_q : quarterly start

Development (from uy_m and cy_m, with dev_m derived if absent):

  • dev_y is the yearly development index, where dev_m 1-12 map to 1, 13-24 map to 2, and so on.

  • dev_h and dev_q are aligned to calendar half-yearly and quarterly boundaries (not simple groupings of dev_m), so cohorts such as Q1 / Q2 / H1 / H2 are compared consistently on the same cumulative development basis.

Newly created columns are inserted before their corresponding base columns.

Examples

if (FALSE) { # \dontrun{
df <- data.frame(
  uy_m  = as.Date("2023-01-01") + 0:5 * 30,
  cy_m  = as.Date("2023-01-01") + 0:5 * 30,
  dev_m = 1:6
)

df2 <- derive_grain_columns(df)
head(df2)
} # }