Interpolate univariate or multivariate time series with missing values

Usage

interpol(incomplete, pattern, extend_range = FALSE)

Arguments

incomplete: a ts-boxable object with NAs
pattern: a ts-boxable object no missing values
extend_range: logical. If TRUE, uses a full_join to potentially extend the date range. If FALSE (default), uses a left_join to only fill NAs within the original date range.

Value

interpolated object of the same type as the input

Details

the function aligns two sets of time series by date and then interpolates the first set using the second set as a pattern. The time-span of the pattern has to match or be larger than the time-span of the incomplete data.

Examples

monthly_data_example |>
  dplyr::filter(time > "2002-01-01") |>
  tsbox::ts_long() |>
  dplyr::mutate(
    value = dplyr::if_else(time > "2021-01-01" & time < "2022-01-01",
    NA_real_, value)
  )|>
  interpol(monthly_data_example, extend_range = TRUE)
#> The mapping betwen interpolated series and pattern series is the following:
#>  VISNS_HI_I  VAPNS_HI_I VADCNS_HI_I 
#>  VISNS_HI_P  VAPNS_HI_P VADCNS_HI_P 
#> # A tibble: 885 × 3
#>    id       time       value
#>    <chr>    <date>     <dbl>
#>  1 VISNS_HI 2000-09-01  539.
#>  2 VISNS_HI 2000-10-01  567.
#>  3 VISNS_HI 2000-11-01  535.
#>  4 VISNS_HI 2000-12-01  584.
#>  5 VISNS_HI 2001-01-01  558.
#>  6 VISNS_HI 2001-02-01  554.
#>  7 VISNS_HI 2001-03-01  618.
#>  8 VISNS_HI 2001-04-01  558.
#>  9 VISNS_HI 2001-05-01  528.
#> 10 VISNS_HI 2001-06-01  597.
#> # ℹ 875 more rows
monthly_data_example |>
  tsbox::ts_long() |>
  dplyr::mutate(
    value = dplyr::if_else(
      time > "2021-01-01" & time < "2022-01-01",
      NA_real_,
      value
    )
  ) |>
  interpol(
    # missing values in the pattern
    monthly_data_example |>
      tsbox::ts_long() |>
      dplyr::mutate(
        value = dplyr::if_else(
          time > "2021-03-01" & time < "2021-09-01",
          NA_real_,
          value
        )
      )
  ) %>%
  # no interpolation where pattern has NAs
  tsbox::ts_wide()
#> The mapping betwen interpolated series and pattern series is the following:
#>  VISNS_HI_I  VAPNS_HI_I VADCNS_HI_I 
#>  VISNS_HI_P  VAPNS_HI_P VADCNS_HI_P 
#> # A tibble: 295 × 4
#>    time       VISNS_HI VAPNS_HI VADCNS_HI
#>    <date>        <dbl>    <dbl>     <dbl>
#>  1 2000-09-01     539.     546.      146.
#>  2 2000-10-01     567.     604.      153.
#>  3 2000-11-01     535.     577.      154.
#>  4 2000-12-01     584.     617.      181.
#>  5 2001-01-01     558.     578.      182.
#>  6 2001-02-01     554.     559.      176.
#>  7 2001-03-01     618.     636.      173.
#>  8 2001-04-01     558.     594.      155.
#>  9 2001-05-01     528.     586.      142.
#> 10 2001-06-01     597.     672.      183.
#> # ℹ 285 more rows