Backfill the daily series of a keyword over a gap

Source: R/proc_keyword_backfill_daily.R

The daily Google Trends download in proc_keyword_latest() only covers the last ~90 days. After an interruption of the daily run, the stored daily series therefore has a hole: everything between the last pre-outage observation and ~90 days ago is missing. proc_combine_freq() uses the daily series as the high-frequency indicator for tempdisagg::td(), which cannot disaggregate across a hole and fails. This function refills the daily series over the gap using the same efficient overlapping windows as proc_keyword_init(), so that the combine and seasonal-adjustment steps work again and the resulting series is fully consistent at daily resolution.

proc_keyword_backfill_daily(
  keyword = "Insolvenz",
  geo = "CH",
  from = NULL,
  wait = 4,
  retry = 6
)

Arguments

keyword

A single keyword for which to process the data.

geo

A character vector denoting the geographic region. Default is "CH".

from

Start date (YYYY-mm-dd) of the daily backfill. Defaults to one month before the last existing daily observation, which comfortably covers the gap.

Value

Invisibly TRUE.

Details

Only the daily series needs backfilling: a normal daily run already refreshes the weekly (one year) and monthly (since 2006) series over their full span.