padding timeseries with sub-second intervals

[blongworth]

I have a dataset recorded at 16Hz with missing rows (gap in timestamps). Can padr be used to pad this to a regular timeseries? This doesn't work:

library(padr)
now <- as.POSIXct("2024-03-01")
x <- data.frame(time = seq(now, now + 1, 1/4), value = 1:5)
y <- x[-3,]
y |> pad()
#> Error: The specified interval is invalid for the datetime variable.
#>   Not all original observation are in the padding.
#>   If you want to pad at this interval, aggregate the data first with thicken.
y |> thicken(1/4)
#> Error in strsplit(interval_string, " "): non-character argument

^{Created on 2024-03-20 with reprex v2.1.0}

There are no ascii by arguments to seq.POSIXt() smaller than second, but supplying a number works for creating sub-second sequences (see above data generation).

[EdwinTh]

The interval of your example data is is second and on this interval there is no observation missing from the data. pad does work for the following example

now <- as.POSIXct("2024-03-01 00:00:03")
x <- data.frame(time = seq(now, as.POSIXct("2024-03-01 00:00:06"), by = "sec"), value = 1:4)
y <- x[-3,]
pad(y)

Thicken does not work because you need to specify a character for the interval, in the example you give a numeric. Moreover, applying thicken to this data has no use because it is already at the lowest interval supported by padr (the second).

[blongworth]

The seq() call from my example produces a sequence from 2024-03-01 00:00:00.00 to 2024-03-01 00:00:01, with five elements 1/4 s apart.

seq(now, now + 1, 1/4)

I was hoping I was missing a way that padr could work with arbitrary intervals, like the 1/4 passed to seq() instead of "sec", "hour", etc.

It sounds like the answer is that padr only supports intervals 1 second or greater.

[EdwinTh]

Yes that conlusion is correct