Linear approximation of missing values

Fill missing values in a time series (or in a column of a longitudinal data set) with the approximation method.

na_approx(indicator)

Arguments

indicator	A tibble created by get_eurostat_indicator. with time, geo, value, frequency, estimate and method columns.

Value

A tibble updated with with approximated values.

See also

Other approximation functions: impute_indicators(), na_locf(), na_nocb()

Examples

{
example_df <- data.frame (
   time = rep(as.Date (paste0(2018:2021, "-01-01")),3),
   geo = c( rep("NL", 4), rep("BE", 4), rep("LU", 4)),
   value = c(10,11, NA_real_, 12, NA_real_, 21,22,23, 5,6,7,NA_real_),
   frequency = rep("A", 12)
)
example_df$method <- ifelse(is.na(example_df$value), "missing", "actual")
example_df$estimate <- example_df$method

na_approx ( example_df )
}
#> # A tibble: 12 x 6
#>    time       frequency geo   value estimate method 
#>    <date>     <chr>     <chr> <dbl> <chr>    <chr>  
#>  1 2018-01-01 A         NL     10   actual   actual 
#>  2 2018-01-01 A         BE     NA   missing  missing
#>  3 2018-01-01 A         LU      5   actual   actual 
#>  4 2019-01-01 A         NL     11   actual   actual 
#>  5 2019-01-01 A         BE     21   actual   actual 
#>  6 2019-01-01 A         LU      6   actual   actual 
#>  7 2020-01-01 A         NL     11.5 approx   approx 
#>  8 2020-01-01 A         BE     22   actual   actual 
#>  9 2020-01-01 A         LU      7   actual   actual 
#> 10 2021-01-01 A         NL     12   actual   actual 
#> 11 2021-01-01 A         BE     23   actual   actual 
#> 12 2021-01-01 A         LU     NA   missing  missing