as.Date converts wrong date from POSIXct data

Question

I have 3848 rows of POSIXct data - stop times of bike trips in the month of April. As you can see, all of the data is in POSIXct format and is within the range of the month of April.

length(output2_stoptime)
[1] 3848
head(output2_stoptime)
[1] "2015-04-01 17:19:27 EST" "2015-04-02 07:26:06 EST" "2015-04-08      10:09:37 EST"
[4] "2015-04-12 20:08:00 EST" "2015-04-13 17:53:11 EST" "2015-04-14    07:17:34 EST"
class(output2_stoptime)
[1] "POSIXct" "POSIXt"
range(output2_stoptime)
[1] "2015-04-01 00:34:29 EST" "2015-04-30 20:49:22 EST"
Sys.timezone()
[1] "EST"

However, when I try converting this into a table of stop times per day, I get 4 dates that are converted as the 1st of May. I thought this might be occurring due to the different system timezone as I am located in Europe at the moment, but even after setting the timezone to EST, the problem persists. For example:

by_day_output2 = as.data.frame(as.Date(output2_stoptime), tz = "EST")
colnames(by_day_output2)[1] = "SUM"

movements_Apr = as.data.frame(table(by_day_output2$SUM))
colnames(movements_Apr)[1] = "DATE"

tail(movements_Apr)
         DATE Freq
26 2015-04-26   96
27 2015-04-27  125
28 2015-04-28  145
29 2015-04-29  151
30 2015-04-30   99 
31 2015-05-01    4

Why are the four dates converting improperly when the time zones of the data and the system match? None of the data falls within May.

`"EST"` is a Canadian timezone, not European. See `?timezone` for how you should specify timezones. — Joshua Ulrich, Jan 07 '16 at 14:59
Sorry, I meant April, not March, I fixed that. The data is New York Citibike data, so EST is the timezone that the system needs to be in for the analysis — iskandarblue, Jan 07 '16 at 15:05
It's so strange, because for example I take a different time like this: `[1] "2015-07-31 23:42:23 EST"` and it is improperly converted as August 1st — iskandarblue, Jan 07 '16 at 15:15
I get the following error message: `Warning message: In format.POSIXlt(as.POSIXlt(x), ...) : unknown timezone 'EDT'` — iskandarblue, Jan 07 '16 at 15:20
And when I convert the system time to EDT I still get the same improper conversion — iskandarblue, Jan 07 '16 at 15:21
Note that this could be an error: `as.data.frame(as.Date(output2_stoptime), tz = "EST")`. Your `tz` argument is specified in the `as.data.frame()` call, not the `as.Date` call. — nrussell, Jan 07 '16 at 15:41
You should still be using the correct *full* time zone identifier, which would be `"America/New_York"` — Matt Johnson-Pint, Jan 07 '16 at 16:59

as.Date converts wrong date from POSIXct data

0 Answers0