how-to convert a list of unix-time values to a time series class object

Question

If I have a vector of unix timestamps, for example

> head(debug$time)
[1] 1.473639e+18 1.473639e+18 1.473639e+18 1.473639e+18 1.473640e+18
[6] 1.473640e+18

How can I use it as a Date vector in order to form a ts or xts object ? I need this in order to parameterise and use the bfast library docs

I tired to do the following (tz=CEST is not recognised btw):

t_ <- as.POSIXct(as.numeric(debug$time), origin = '1970-01-01')

Could you please advise on this one ?

Where did those #'s come from? They're about 1,000,000,000 times too large. — hrbrmstr, Oct 01 '16 at 18:41
@hrbrmstr is in microseconds (1/1,000,000 second). oops, you are right: 10^9 smaller — nskalis, Oct 01 '16 at 18:47

score 2 · Accepted Answer · edited May 23 '17 at 12:33

2

Use this for a way around the CEST problem, but all you need to do is:

vals <- c(1.473639e+18, 1.473639e+18, 1.473639e+18, 1.473639e+18, 1.473640e+18, 1.473640e+18)
as.POSIXct(vals/1000000000, origin="1970-01-01 00:00:00")
## [1] "2016-09-11 20:10:00 EDT" "2016-09-11 20:10:00 EDT" "2016-09-11 20:10:00 EDT" "2016-09-11 20:10:00 EDT" "2016-09-11 20:26:40 EDT"
## [6] "2016-09-11 20:26:40 EDT"

To make the date/time conversion work.

UPDATE

Looks like you can also do:

as.POSIXct(vals/1e9, origin="1970-01-01 00:00:00", tz="Etc/GMT+2")

to get your CEST working as well (from what I read, that's an equivalent time zone, but I don't live there or work with data from there so cannot validate that statement). Also using the scientific notation for Rich's sake :-)

edited May 23 '17 at 12:33

Community

1
1

answered Oct 01 '16 at 18:49

hrbrmstr

77,368
11
139
205

`1e9` is 1000000000, but you know that ;) Fore! – Rich Scriven Oct 01 '16 at 18:50
aye. i kept both in long notation for effect. especially since that level of precision was not needed for these numbers. :-) – hrbrmstr Oct 01 '16 at 18:53

how-to convert a list of unix-time values to a time series class object

1 Answers1