Time formats

How to convert your time to a format that R can understand as time.

Miriam Lerma true
2022-06-30

Intro

This post is about how to convert date and time to a format that R can understand it.

For example if you wrote the date and time during fieldwork in a notebook, and then wrote it down in a spreadsheet at your computer. When you load your this spreadsheet into R, very often R would think that your date and time is text. Therefore, you need to change the ‘class’ of the column where you have date and time.

It is important to change the ‘class’ of your column, so you can:
- Correct the time zone
- Calculate duration of events (for example foraging trip duration)
- Count days (for example Julian days)
- Identify night and day periods
… and many many more

In this post we will transform the class of our column date and time from character or other class, into POSTIXct. Additionally, we will change the time zone.

Data 📖

To do this exercise, load data from the package ‘sula’.
For accessing the data, you need to have the package installed.

To install:

#devtools::install_github("MiriamLL/sula")
library(sula)

The data is from 10 tracked individuals.

GPS_raw<-(GPS_raw)

Date 📅

The data frame contains a column with date information.

head(GPS_raw$DateGMT)
[1] "02/11/2017" "02/11/2017" "02/11/2017" "02/11/2017" "02/11/2017"
[6] "02/11/2017"

The date infomation is of class character.

class(GPS_raw$DateGMT)
[1] "character"

Time ⏰

The data frame contains a column with time information.

head(GPS_raw$TimeGMT)
[1] "17:05:30" "17:09:35" "17:13:50" "17:17:59" "17:22:13" "17:26:25"

The time infomation is of class character.

class(GPS_raw$TimeGMT)
[1] "character"

Transform the column to POSIXct

POSIX comes from Portable Operating System Interface and the X from UNIX.
There are two subclasses ct from calendar time and lt from local time. See R news.
Calendar time being the number of seconds since the beginning of 1970.


To create our column with date and time, lets merge the column DateGMT and TimeGMT from our data frame into one column DateTime.

GPS_raw$DateTime<-paste(GPS_raw$DateGMT,GPS_raw$TimeGMT)

Then convert it to POSIXct

GPS_raw$DateTime<-as.POSIXct(strptime(GPS_raw$DateTime, format = "%d/%m/%Y %H:%M:%S"))

Arguments used:

strptime is a function to convert characters to time objects, where str stands for string, p from parse, and time is self explanatory.

Then we provide the information of the format we were using, be specially careful with the format of your date and time.

For the example, I used %d/%m/%Y %H:%M:%S

DAY
%d - is used for day and goes from 0 to 31
%a - is used for the day of the week using the first three letters of the name
%A - is used for the day of the week using the full name

MONTH
%m - is used for month it goes from 0 to 12
%b - is used for the name of the month using the first three letters of the name
%B - is used for the name of the month using the full name

YEAR
%y - is used for year using two digits
%Y - is used for year using four difits

HOURS
%H - is used for hours and goes from 0 to 24
MINUTES
%M is used for minutes and goes from 0 to 60
SECONDS
%S is used for seconds and goes from 0 to 60

It is also important to note if you are using slash (/), dash (-) or any other, and do not forget the colon (:) when separating the times

Note for excel-users: if you open your file first in excel, excel tries to identify the class, and might have transform the column. Therefore you might need to use one of the examples from stackoverflow

Check

Many struggles come from the format. If the format you are giving doesnt corresponds to the format of your column, it would generate you NAs.
Therefore, I recommend to create a column to make the trials on the format and to always check that you have the correct transformations.

Here is an example for checking:

range(GPS_raw$DateTime)
[1] "2017-11-02 17:05:30 CET" "2017-11-27 09:48:57 CET"

Note that in this example it returns the word CET.

This is because I am in the Central European Time.

This might be correct and all good, but it is useful to know that you can transform it to you time zone.

Correct time 🌐

One way to directly have the format in our time zone, is to specify the time zone when converting the class.

Here, I am adding the “GMT” at the end.

GPS_raw$GMT<-as.POSIXct(strptime(GPS_raw$DateTime, format = "%Y-%m-%d %H:%M:%S"),"GMT")

And now shows GMT.

range(GPS_raw$GMT)
[1] "2017-11-02 17:05:30 GMT" "2017-11-27 09:48:57 GMT"

Identify your time zone 🗺️

According to where your study was made, you might need to change the time to your time zone tz

There are many ways to identify your tz:

In case you are here because you are interested in analyzing tracking data, note that it’s common that the GPS record data in GMT+0.

lubridate

One option to change your date and time to your time zone is using the package lubridate

library(lubridate)

Example with the data.

GPS_tz<-GPS_raw

My original tz:

GPS_tz$CET<-ymd_hms(GPS_tz$DateTime, tz = "Europe/Amsterdam")

My goal tz:

GPS_tz$UTC_4<- with_tz(GPS_tz$CET,tzone = "America/La_Paz")

Manually

If you know the time difference between the recording and the region where you are, you can also calculate it manually.

For example:

GPS_tz$five_hours_difference <- GPS_tz$CET - 3600*5

Thats it!

Hopefully this would help you.

Recommendations

Be very careful with the format and your time zone.

I would recommend that you always create an extra column in your data frame to make the transformations, not your original column because if it returns NAs you will have to load the data frame over and over.

It takes time to get use to this transformations and there are many different ways to transform times and date, so if you are struggling, you are not the only one, just give it some time.

Other resources 👩🏽‍🏫