Fill Imputation

Imputes a new vector of variables which can be used to fill NA values.

Usage

impute_fill(ind, var, times, FUN_n)

Arguments

ind

index vector

var

numeric vector of values containing NA_real values for imputation

times

POSIXct vector of time in hourly increments

FUN_n

rolling window size

Value

vector of var values filling NA

Examples

dt <- weather
rolling_window <- length(dt[is.na(temp),temp])*2
dt[,indx := .I]
#> Index: <station>
#>                     times      temp    rh  rain    ws    wd        wd_sd    lon
#>                    <POSc>     <num> <num> <num> <num> <num>        <num>  <num>
#>    1: 2020-04-13 01:00:00 11.710215   NaN   0.2   1.9   303 0.000000e+00 135.86
#>    2: 2020-04-13 02:00:00  8.591594   NaN   0.0   1.0    38 0.000000e+00 135.86
#>    3: 2020-04-13 03:00:00  5.029886   NaN   0.0   3.5    15 0.000000e+00 135.86
#>    4: 2020-04-13 04:00:00  7.329537   NaN   0.0   0.4   194 8.537737e-07 135.86
#>    5: 2020-04-13 05:00:00  4.867557   NaN   0.4   1.9   276 0.000000e+00 135.86
#>   ---                                                                          
#> 8782: 2020-10-12 21:00:00 14.505099   NaN   0.0   1.6   189 0.000000e+00 135.73
#> 8783: 2020-10-12 22:00:00 16.143942   NaN   0.0   1.2   163 0.000000e+00 135.73
#> 8784: 2020-10-12 23:00:00 10.410922   NaN   0.0   3.4   137 0.000000e+00 135.73
#> 8785: 2020-10-13 00:00:00  7.722190   NaN   0.0   3.3   117 0.000000e+00 135.73
#> 8786: 2020-10-13 01:00:00  5.800267   NaN   0.0   3.6   112 0.000000e+00 135.73
#>          lat station  YYYY    MM    DD    hh    mm  indx
#>        <num>  <char> <num> <num> <int> <int> <num> <int>
#>    1: -33.26 naddacs  2020     4    13     1     0     1
#>    2: -33.26 naddacs  2020     4    13     2     0     2
#>    3: -33.26 naddacs  2020     4    13     3     0     3
#>    4: -33.26 naddacs  2020     4    13     4     0     4
#>    5: -33.26 naddacs  2020     4    13     5     0     5
#>   ---                                                   
#> 8782: -33.13 scaddan  2020    10    12    21     0  8782
#> 8783: -33.13 scaddan  2020    10    12    22     0  8783
#> 8784: -33.13 scaddan  2020    10    12    23     0  8784
#> 8785: -33.13 scaddan  2020    10    13     0     0  8785
#> 8786: -33.13 scaddan  2020    10    13     1     0  8786

dt[, tm_imp := round(data.table::frollapply(
  indx,
  n = rolling_window,
  fill = NA_real_,
  FUN = impute_fill,
  FUN_n = rolling_window,
  times = times,
  var = temp,
  align = "center"
  ),3)]
#> Index: <station>
#>                     times      temp    rh  rain    ws    wd        wd_sd    lon
#>                    <POSc>     <num> <num> <num> <num> <num>        <num>  <num>
#>    1: 2020-04-13 01:00:00 11.710215   NaN   0.2   1.9   303 0.000000e+00 135.86
#>    2: 2020-04-13 02:00:00  8.591594   NaN   0.0   1.0    38 0.000000e+00 135.86
#>    3: 2020-04-13 03:00:00  5.029886   NaN   0.0   3.5    15 0.000000e+00 135.86
#>    4: 2020-04-13 04:00:00  7.329537   NaN   0.0   0.4   194 8.537737e-07 135.86
#>    5: 2020-04-13 05:00:00  4.867557   NaN   0.4   1.9   276 0.000000e+00 135.86
#>   ---                                                                          
#> 8782: 2020-10-12 21:00:00 14.505099   NaN   0.0   1.6   189 0.000000e+00 135.73
#> 8783: 2020-10-12 22:00:00 16.143942   NaN   0.0   1.2   163 0.000000e+00 135.73
#> 8784: 2020-10-12 23:00:00 10.410922   NaN   0.0   3.4   137 0.000000e+00 135.73
#> 8785: 2020-10-13 00:00:00  7.722190   NaN   0.0   3.3   117 0.000000e+00 135.73
#> 8786: 2020-10-13 01:00:00  5.800267   NaN   0.0   3.6   112 0.000000e+00 135.73
#>          lat station  YYYY    MM    DD    hh    mm  indx tm_imp
#>        <num>  <char> <num> <num> <int> <int> <num> <int>  <num>
#>    1: -33.26 naddacs  2020     4    13     1     0     1     NA
#>    2: -33.26 naddacs  2020     4    13     2     0     2     NA
#>    3: -33.26 naddacs  2020     4    13     3     0     3     NA
#>    4: -33.26 naddacs  2020     4    13     4     0     4     NA
#>    5: -33.26 naddacs  2020     4    13     5     0     5     NA
#>   ---                                                          
#> 8782: -33.13 scaddan  2020    10    12    21     0  8782     NA
#> 8783: -33.13 scaddan  2020    10    12    22     0  8783     NA
#> 8784: -33.13 scaddan  2020    10    12    23     0  8784     NA
#> 8785: -33.13 scaddan  2020    10    13     0     0  8785     NA
#> 8786: -33.13 scaddan  2020    10    13     1     0  8786     NA

plot(dt[2950:3100, temp], type = "l")
lines(dt[2950:3100, tm_imp], type = "l", col = "red")