for working with financial time series, like daily stock prices or intraday data, which time series packages are preferred? xts, plain zoo, or timeSeries or something else? I use both xts and zoo, but sometimes not sure to use xts exclusively or sometimes zoo have advantage of lighter overhead; also, I remembered a review paper on all these packages by Rmetrics, which claims that xts cannot even finish some tests they did. But I cannot find the paper now.
-
The rmetrics paper is on the rmetrics website. – Shane May 28 '10 at 19:58
-
Related: [Which R time/date class and package to use?](http://stackoverflow.com/questions/4354974/which-r-time-date-class-and-package-to-use) – Joshua Ulrich Aug 13 '11 at 18:06
2 Answers
I am rather happy with xts and zoo and alternate between the two.
Nothing forces you to use one exclusively. As zoo is a little older, some packages interface it rather than xts. But xts has extensions which provide extra functionality (e.g. the indexing) which make it a valid option.
Other folks may be perfectly happy with Rmetrics classes. It all depends, and is to some extent a matter of personal preferences.

- 360,940
- 56
- 644
- 725
I myself use zoo as my 'default' option since I have the feeling this is the most common time series class used in the community.
I think both xts and zoo are good choices. SInce they are well known in the community.
There are even more time series classes available apart from xts and zoo. See also Task View Time Series. But I myself try to stay away from too exotic time series classes except I need the particular features. This makes it easier for others to understand the code.

- 7,277
- 11
- 36
- 55