[R] ts & daily timeseries
v.demartino2 at virgilio.it
Wed Jun 23 15:33:52 CEST 2004
On Tuesday 22 June 2004 17:41, Prof Brian Ripley wrote:
> data.frame(time=mytime, ts=myts)
> would appear to be what you are looking for.
> On Tue, 22 Jun 2004 v.demartino2 at virgilio.it wrote:
> > I have defined a daily timeseries for the 365 days of 2003 issuing:
> > myts = ts(dati[,2:10],frequency=365,)
> > > myts
> > Time Series:
> > Start = c(1, 1)
> > End = c(1, 365)
> > Frequency = 365
> > and
> > mytime = as.POSIXct(strptime(as.character(dati[,1]),format="%Y-%m-%d"))
> > contains the dates from "2003-01-01" to "2003-12-31"
> Why not use the Date class? However, if all you want is the character
> representation of the dates you appear to have those in dati.
> > How can I combine mytime and myts in order to list the timeseries
> > according to the more natural
> > 2003-01-01 xxxxx xxxxx xxxxx ......
> > 2003-01-02 xxxxx xxxxx xxxxx ......
> > ...........
> > 2003-12-31 xxxxx xxxxx xxxxx ......
> What do you want the first column to be? Character strings? Dates? What
> do you want to do with this?
As a newbye becoming more and more confident with R, I'm having a go at neural
networks by means of your nnet package trying to model daily electricity
demand with respect to time, weather extreme temperatures,air humidity, sky
coverage (don't know if is correct in English, I mean: a measurement of the
intensity of daylight going to very cloudy to completely clear), and type of
day (working days, Mon, Sat, Sun).
Therefore, among many other things, I need to:
1) Window different spells of the year to study them separatedly;
2) Print in a human readable way the list of data of the spell under
3) Plot the same spells as timeseries, that is with a time referenced x-axis.
More information about the R-help