R：每日数据到每月

Question

I have a large xts object, with multiple variable. The index is daily in that manner, it corresponds to exact days, however there is only one observation for each variable in a month. Is there a way to drop the day from the index and only keep year-month?

To ilustrate my problem for instance I have var1 with an observation on 2011-06-28 and var2 with observation 2011-06-30. I would like to index both as 2011-06

Thanks

Answer 1

You can probably do this: Use gsub (replace a pattern with whatever you want) with regex (a sequence of characters that define a search pattern in eg a string). The pattern is done with regex, which has lots of metacharacters that allow you to do more advanced things. The dot (.) is a wildcard and the $ anchors it at the back. So the pattern is basically any 3 characters before the end and replace them with nothing.

your_object<-c("2011-06-28","2011-06-30")
gsub(pattern = "...$", replace = "", x = your_object)

Here is a guide for using gsub with regex ( http://uc-r.github.io/regex ).

Answer 2

alternatively you could "tell" R that you are using dates of a certain format with the as.Date() function and then use format() to change it to the format you desire. Like this:

dates=c("2011-06-28","2011-06-29","2011-06-30","2011-07-1") #test string with dates in original format

dates2 <- format(as.Date(dates,"%Y-%m-%d"), format="%Y-%m") #changing the "%Y-%m-%d" format to  the desired "%Y-%m"

print(dates2) 

Edit: If you only want to change the index of a xts: indexFormat(xts_object) <- "%Y-%m"

Cheers Chris