[R] How to group a count

[Deepayan Sarkar]

On 5/15/07, Spilak,Jacqueline [Edm] <Jacqueline.Spilak at ec.gc.ca> wrote:
> Here is my data.  I tried table but it doesn't do what I want it to do
> when it graphs.  I want a count of the types (R for one graph and A for
> another) by hour grouped by year.  Hope that helps.
>
> ID,YYYY,MM,DD,HH,MM,Type
> YEG,2002,01,01,01,24,A
> YEG,2002,01,01,02,40,R
> YEG,2002,01,01,05,39,R
> YEG,2002,01,01,09,17,A

I assume you have more data than that. Here's a sample use of xtabs,
which you should adapt to your example (I'm not sure if you want to
disregard MM and DD):

> foo
   ID YYYY MM DD HH MM.1 Type
1 YEG 2002  1  1  1   24    A
2 YEG 2002  1  1  2   40    R
3 YEG 2002  1  1  5   39    R
4 YEG 2002  1  1  9   17    A
> as.data.frame(xtabs(~Type + YYYY + HH, foo))
  Type YYYY HH Freq
1    A 2002  1    1
2    R 2002  1    0
3    A 2002  2    0
4    R 2002  2    1
5    A 2002  5    0
6    R 2002  5    1
7    A 2002  9    1
8    R 2002  9    0

(xtabs itself will produce a 3-D array, which you may or may not be
comfortable working with).

-Deepayan

>
> -----Original Message-----
> From: Deepayan Sarkar [mailto:deepayan.sarkar at gmail.com]
> Sent: May 15, 2007 3:46 PM
> To: Spilak,Jacqueline [Edm]
> Cc: r-help at stat.math.ethz.ch
> Subject: Re: [R] How to group a count
>
> On 5/15/07, Spilak,Jacqueline [Edm] <Jacqueline.Spilak at ec.gc.ca> wrote:
> > Hello R users
> > I have a dataset that has different types of records with different
> > dates and times pertaining to each.  I would like to have a bar graph
> > of a count of the types(ie. The number of types) of recods by hour
> > grouped by year.  So the count of the types would be the y axis, the
> > hour on the x axis and then grouped by year for easy comparison.  I
> > think that I have to use barchart however I don't know how to get
> > barchart to do a count and not graph values.
>
> I think you want to use table or xtabs to get a frequency table, and use
> barchart on the result. Hard to say more without an example.
>
> -Deepayan
>