[R] Newbie: Better way to do compound conditionals in subset?

[Marc Schwartz]

On Thu, 2006-10-26 at 15:55 -0400, Zembower, Kevin wrote:
> There must be a better way to select the rows after 22-Apr-2004 and
> before 01-Sep-2004 with a temperature below 65 than this:
> 
> > before2sw1 <- subset(energy.data, as.Date(start, format="%d-%b-%y") <
> as.Date("01-Sep-04", format = "%d-%b-%y"))
> > before2sw2 <- subset(before2sw1, as.Date(start, format="%d-%b-%y") >=
> as.Date("22-Apr-04", format = "%d-%b-%y"), select=c(therms,temp,days))
> > before2sw <- subset(before2sw2, temp < 65)

Well, the first thing that I would do is to convert 'start' to a Date:

  energy.data$start <- as.Date(energy.data$start, format = "%d-%b-%y")

Then create your cutoff dates:

  Start <- as.Date("22-Apr-04", format = "%d-%b-%y")
  End <- as.Date("01-Sep-04", format = "%d-%b-%y")

Then you can do this:

  New.DF <- subset(energy.data, 
                  (start >= Start) & (start < End) & (days < 65),
                  select = c(therms, temp, days))

and then add:

  NewDF$HDD <- with(NewDF, (therms / (65 - temp) * days))

You can also do the calculations and add a new column to the full data
frame and simply add 'HDD' to the 'select' argument in subset().

Note that if you don't want to modify the original data frame, which is
something that I tend to avoid for a variety of reasons, you can copy it
to another first and then run the above steps on the copy for subsequent
analysis.

You want to generally avoid using attach(), as it can have deleterious
and not immediately evident side-effects.  Review the Details section
of ?attach and note some of the effects seen in the Examples there.

HTH,

Marc Schwartz

> Is it also possible to combine in this step:
> 
> attach(before2sw)
> before2sw.HDD <- therms / (65 - temp) * days
> 
> My data looks like this:
> > head(energy.data)
>       start therms   gas KWHs elect temp days
> 1 10-Jun-98      9 16.84  613 63.80   75   40
> 2 20-Jul-98      6 15.29  721 74.21   76   29
> 3 18-Aug-98      7 15.73  597 62.22   76   29
> 4 16-Sep-98     42 35.81  460 43.98   70   33
> 5 19-Oct-98    105 77.28  314 31.45   57   29
> 6 17-Nov-98    106 77.01  342 33.86   48   30
> >
> 
> Thanks for your suggestions and advice. I'm continuing to enjoy learning
> R.
> 
> -Kevin