Fill missing values in a FRED series

Fills NA values in the value column (long format) or in numeric columns (wide format). Two methods are supported: last-observation-carry-forward ("locf") and linear interpolation between adjacent observed values ("linear"). Use this for mixed-frequency analysis where a low-frequency series needs to be interpolated to a higher frequency.

Usage

fred_interpolate(data, method = c("locf", "linear"))

Arguments

data

A fred_tbl or data.frame with a date column.

method

Character. "locf" (default) carries the last observed value forward. "linear" does linear interpolation between adjacent non-NA values.

Value

A fred_tbl with interior NAs filled (see boundary note above).

Details

Boundary behaviour: with method = "locf", leading NAs remain NA because there is no prior observation to carry forward. With method = "linear", neither leading nor trailing NAs are filled because stats::approx() is called with rule = 1 (no extrapolation). If you need extrapolation, post-process the result.

See also

Other utilities: fred_aggregate(), fred_event_window()

Examples

# Synthetic example: fill interior NAs
d <- seq(as.Date("2024-01-01"), by = "month", length.out = 6L)
df <- data.frame(date = d, series_id = "X",
                 value = c(NA, 2, NA, NA, 5, NA))
fred_interpolate(df, method = "locf")
#> # FRED: 6 rows
#>         date series_id value
#> 1 2024-01-01         X    NA
#> 2 2024-02-01         X     2
#> 3 2024-03-01         X     2
#> 4 2024-04-01         X     2
#> 5 2024-05-01         X     5
#> 6 2024-06-01         X     5
fred_interpolate(df, method = "linear")
#> # FRED: 6 rows
#>         date series_id value
#> 1 2024-01-01         X    NA
#> 2 2024-02-01         X     2
#> 3 2024-03-01         X     3
#> 4 2024-04-01         X     4
#> 5 2024-05-01         X     5
#> 6 2024-06-01         X    NA

# \donttest{
op <- options(fred.cache_dir = tempdir())
if (FALSE) { # \dontrun{
  gdp <- fred_series("GDPC1", from = "2020-01-01")
  gdp_monthly <- fred_interpolate(gdp, method = "linear")
} # }
options(op)
# }