Aggregate a daily series to weeks

Aggregates daily rows to calendar weeks, taking the mean of some columns (stocks) and the sum of others (flows).

Usage

to_weekly(
  dt,
  by = NULL,
  date_col = "date",
  mean_cols = NULL,
  sum_cols = NULL,
  week_start = 1L
)

Arguments

dt

A data.frame or data.table with one row per by x date.

by

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

date_col

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

mean_cols

Character vector of numeric columns averaged over the week, or NULL.

sum_cols

Character vector of numeric columns summed over the week, or NULL.

week_start

First day of the week, integer in 1..7 with 1 = Monday and 7 = Sunday.

Value

A data.table with columns by, week (IDate), n_days (integer), then mean_cols and sum_cols under their own names, ordered by by and week.

Details

Weeks start on week_start (ISO numbering: 1 = Monday, ..., 7 = Sunday; default Monday, as in ISO 8601 weeks). The week column holds the first day of the week. Partial weeks (at the edges of the series or with missing days) are kept: n_days reports the number of days present, so the caller can drop weeks with n_days < 7. NA values propagate to the weekly mean or sum.

Examples

d <- data.table::data.table(
  date = data.table::as.IDate("2024-01-01") + 0:13,
  stock = 100:113,
  inflow = rep(1:2, 7)
)
to_weekly(d, mean_cols = "stock", sum_cols = "inflow")
#>          week n_days stock inflow
#>        <IDat>  <int> <num>  <int>
#> 1: 2024-01-01      7   103     10
#> 2: 2024-01-08      7   110     11