The R Book

Dates in Dataframes

There is an introduction to the complexities of using dates and times in dataframes on pp. 89–95. Here we work with a simple example:

nums<-read.table("c:\\temp\\sortdata.txt",header=T)
attach(nums)
names(nums)

[1]  "name"  "date"  "response"  "treatment"

The idea is to order the rows by date. The ordering is to be applied to all four columns of the dataframe. Note that ordering on the basis of our variable called date does not work in the way we want it to:

nums[order(date),]

         name         date       response   treatment
53      rachel  01/08/2003    32.98792196           B
65      albert  02/06/2003    38.41979568           A
6         ann   02/07/2003     2.86983693           B
10      cecily  02/11/2003     6.81467571           A
4         ian   02/12/2003     2.09505949           A
29    michael   03/05/2003    15.59890900           B
...

This is because of the format used for depicting date in the dataframe called nums: date is a character string in which the first characters are the day, then the month, then the year. When we sort by date, we typically want 2001 to come before 2006, May 2006 before September 2006 and 12 May 2006 before 14 May 2006. In order to sort by date we need first to convert our variable into date-time format using the strptime function (see p. 92 for details):

dates<-strptime(date,format="%d/%m/%Y")
dates

[1]   "2003-08-25"   "2003-05-21"   "2003-10-12"   "2003-12-02"  "2003-10-18"
[6]   "2003-07-02"   "2003-09-27"   "2003-06-05"   "2003-06-11"  "2003-11-02"

Note how strptime has produced a date object with year first, then a hyphen, then month, then a hyphen, then day which will sort ...