R help - Feb 2002 - A couple of little R things I can't figure out (column percents, regression with lagged variables)

Simple usage questions that I ought to be able to figure on my own, but 
can't.

1.I'm able to produce a cross tabulation table showing counts with 
either table or xtabs.  But I want column percentages for 
interpretation, and it seems stupid to sit there with a calculator 
figuring marginals and column percentages.  How to make R do it after this:

 > x <- c(1,3,1,3,1,3,1,3,4,4)
 > y <- c(2,4,1,4,2,4,1,4,2,4)
 > hmm <- table(x,y)
 > hmm
    y
x   1 2 4
   1 2 2 0
   3 0 0 4
   4 0 1 1

#I can get the column sums:
 > tots <- apply(hmm,2,sum)

and I can get the total N of counts, but don't understand how to make it 
calculate column percents, as in

     y

x   1    2     4
   1 100 66.67  0
   3  0   0     80
   4  0   33.33 20

Pointers appreciated.


2. A student said here's  y, a vector representing a time series, and 
here's x, a vector representing a time series. I want to do a 
conventional regression of y on the lag of x. In sas you do xlag=lag(x) 
and then use xlag in a regresson.  I just want something simple like 
lm(y~lag(x)). But in R base there's no lag.

So I can get it the old fashioned way:
 > xx <- c(NA,x)
 > modl <- lm(y~xx[1:length(y)])
 > summary(modl)

One sidenote is that summary does not include any mention of the fact 
that 1 observation was lost due to missing value. That seems bad to me.

I see the lag function in ts, but when I use it, it doesn't change x, so 
obviously I don't see the point of that.

 > z <- lag (x)
 > z
  [1] 1 3 1 3 1 3 1 3 4 4


So long, thanks in advance, greetings, etc...

-- 
Paul E. Johnson                       email: pauljohn at ukans.edu
Dept. of Political Science            http://lark.cc.ku.edu/~pauljohn
University of Kansas                  Office: (785) 864-9086
Lawrence, Kansas 66045                FAX: (785) 864-5700

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Paul Johnson <pauljohn at ku.edu> writes:
>      y
> 
> x   1    2     4
>    1 100 66.67  0
>    3  0   0     80
>    4  0   33.33 20
> 
> Pointers appreciated.
prop.table(hmm,2) * 100
> 2. A student said here's  y, a vector representing a time series, and
> here's x, a vector representing a time series. I want to do a
> conventional regression of y on the lag of x. In sas you do
> xlag=lag(x) and then use xlag in a regresson.  I just want something
> simple like lm(y~lag(x)). But in R base there's no lag.
> 
> So I can get it the old fashioned way:
>  > xx <- c(NA,x)
>  > modl <- lm(y~xx[1:length(y)])
>  > summary(modl)
> 
> One sidenote is that summary does not include any mention of the fact
> that 1 observation was lost due to missing value. That seems bad to me.
> 
> I see the lag function in ts, but when I use it, it doesn't change x,
> so obviously I don't see the point of that.
> 
>  > z <- lag (x)
>  > z
>   [1] 1 3 1 3 1 3 1 3 4 4
This stuff only works for time series, e.g. cbind(ts(x),lag(ts(x))).
Notice that even then, lag() works in the opposite direction (for
historical reasons).

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Try with prop.table(hmm,#)

put # = 1 for row percentages # = 2 for column ones

Stefano

-- Messaggio Originale --
>Simple usage questions that I ought to be able to figure on my own, but
>can't.
>
>1.I'm able to produce a cross tabulation table showing counts with 
>either table or xtabs.  But I want column percentages for 
>interpretation, and it seems stupid to sit there with a calculator 
>figuring marginals and column percentages.  How to make R do it after this:
>
> > x <- c(1,3,1,3,1,3,1,3,4,4)
> > y <- c(2,4,1,4,2,4,1,4,2,4)
> > hmm <- table(x,y)
> > hmm
>    y
>x   1 2 4
>   1 2 2 0
>   3 0 0 4
>   4 0 1 1
>
>#I can get the column sums:
> > tots <- apply(hmm,2,sum)
>
>and I can get the total N of counts, but don't understand how to make it
>
>calculate column percents, as in
>
>     y
>
>x   1    2     4
>   1 100 66.67  0
>   3  0   0     80
>   4  0   33.33 20
>
>Pointers appreciated.
>
>
>2. A student said here's  y, a vector representing a time series, and 
>here's x, a vector representing a time series. I want to do a 
>conventional regression of y on the lag of x. In sas you do xlag=lag(x)
>and then use xlag in a regresson.  I just want something simple like 
>lm(y~lag(x)). But in R base there's no lag.
>
>So I can get it the old fashioned way:
> > xx <- c(NA,x)
> > modl <- lm(y~xx[1:length(y)])
> > summary(modl)
>
>One sidenote is that summary does not include any mention of the fact 
>that 1 observation was lost due to missing value. That seems bad to me.
>
>I see the lag function in ts, but when I use it, it doesn't change x, so
>
>obviously I don't see the point of that.
>
> > z <- lag (x)
> > z
>  [1] 1 3 1 3 1 3 1 3 4 4
>
>
>So long, thanks in advance, greetings, etc...
>
>-- 
>Paul E. Johnson                       email: pauljohn at ukans.edu
>Dept. of Political Science            http://lark.cc.ku.edu/~pauljohn
>University of Kansas                  Office: (785) 864-9086
>Lawrence, Kansas 66045                FAX: (785) 864-5700
>
>-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
>r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
>Send "info", "help", or "[un]subscribe"
>(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
>_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
>


__________________________________________________________________
Abbonati a Tiscali!
Con Tiscali By Phone puoi anche ascoltare ed inviare email al telefono.
Chiama Tiscali By Phone all' 892 800        http://byphone.tiscali.it




-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

hallo paul,

for the percentages use this code (the trick is to use transpose twice)

t(t(hm)/apply(hmm,2,sum))*100




Paul Johnson wrote:
> 
> Simple usage questions that I ought to be able to figure on my own, but
> can't.
> 
> 1.I'm able to produce a cross tabulation table showing counts with
> either table or xtabs.  But I want column percentages for
> interpretation, and it seems stupid to sit there with a calculator
> figuring marginals and column percentages.  How to make R do it after this:
> 
>  > x <- c(1,3,1,3,1,3,1,3,4,4)
>  > y <- c(2,4,1,4,2,4,1,4,2,4)
>  > hmm <- table(x,y)
>  > hmm
>     y
> x   1 2 4
>    1 2 2 0
>    3 0 0 4
>    4 0 1 1
> 
> #I can get the column sums:
>  > tots <- apply(hmm,2,sum)
> 
> and I can get the total N of counts, but don't understand how to make
it
> calculate column percents, as in
> 
>      y
> 
> x   1    2     4
>    1 100 66.67  0
>    3  0   0     80
>    4  0   33.33 20
> 
> Pointers appreciated.
> 
> 2. A student said here's  y, a vector representing a time series, and
> here's x, a vector representing a time series. I want to do a
> conventional regression of y on the lag of x. In sas you do xlag=lag(x)
> and then use xlag in a regresson.  I just want something simple like
> lm(y~lag(x)). But in R base there's no lag.
> 
> So I can get it the old fashioned way:
>  > xx <- c(NA,x)
>  > modl <- lm(y~xx[1:length(y)])
>  > summary(modl)
> 
> One sidenote is that summary does not include any mention of the fact
> that 1 observation was lost due to missing value. That seems bad to me.
> 
> I see the lag function in ts, but when I use it, it doesn't change x,
so
> obviously I don't see the point of that.
> 
>  > z <- lag (x)
>  > z
>   [1] 1 3 1 3 1 3 1 3 4 4
> 
> So long, thanks in advance, greetings, etc...
> 
> --
> Paul E. Johnson                       email: pauljohn at ukans.edu
> Dept. of Political Science            http://lark.cc.ku.edu/~pauljohn
> University of Kansas                  Office: (785) 864-9086
> Lawrence, Kansas 66045                FAX: (785) 864-5700
> 
>
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
> r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
> Send "info", "help", or "[un]subscribe"
> (in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
>
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
-- 
Joerg Maeder    .:|:||:..:.||.::   maeder at atmos.umnw.ethz.ch
Tel: +41 1 633 36 25   .:|:||:..:.||.::   
http://www.iac.ethz.ch/staff/maeder
PhD student at INSTITUTE FOR ATMOSPHERIC AND CLIMATE SCIENCE (IACETH)
ETH Z?RICH Switzerland
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Long one - sorry in advance.

On Tue, Feb 12, 2002 at 12:37:55AM -0600, Paul Johnson
wrote:
> 1.I'm able to produce a cross tabulation table showing counts with 
> either table or xtabs.  But I want column percentages for ...
apply(hmm,2,function(zz){zz/sum(zz)}) * 100

> 2. A student said here's  y, a vector representing a time series, and 
> here's x, a vector representing a time series. I want to do a 
> conventional regression of y on the lag of x. In sas you do xlag=lag(x) 
...
> I see the lag function in ts, but when I use it, it doesn't change x,
so
> obviously I don't see the point of that.
In R, lag(x) shifts the "start" attribute of the time series, but
doesn't alter the data per se.  This is why you get the same before
and after picture, when you just look at it as a vector.  

To see the effect of lagging data, it only makes sense to  compare 
one lagged time series with another time series, like so...

(using x and y above)
> library(ts)
> yt<-ts(y)
> xt<-ts(x)
> lagxt<-lag(xt)
> x1<-ts.union(yt,xt,lagxt)
> x1
Time Series:
Start = 0 
End = 10 
Frequency = 1 
   yt xt lagxt
 0 NA NA     1
 1  2  1     3
 2  4  3     1
 3  1  1     3
 4  4  3     1
 5  2  1     3
 6  4  3     1
 7  1  1     3
 8  4  3     4
 9  2  4     4
 10  4  4    NA

Note the "start" value of x1.

If you want to use lm and friends, you must make them
"plain vanilla" vectors - the time series part confuses
lm.  One way...
> d1<-data.frame(x1)
> r1<-lm(yt ~ lagxt, data=d1)

...
> One sidenote is that summary does not include any mention of the fact 
> that 1 observation was lost due to missing value. That seems bad to me.
...

It does 
> summary(r1)
..[lots of stuff removed]...

Residual standard error: 1.171 on 7 degrees of freedom
Multiple R-Squared: 0.3143,	Adjusted R-squared: 0.2163 
F-statistic: 3.208 on 1 and 7 DF,  p-value: 0.1164 
                      ^^^^^^^^^^

though I'll be the first to admit it doesn't jump up and 
bite you  ;)

Cheers

Jason
-- 
Indigo Industrial Controls Ltd.
64-21-343-545
jasont at indigoindustrial.co.nz
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

At 11:14 PM 2/11/02, Peter Dalgaard BSA wrote:
 >Paul Johnson <pauljohn at ku.edu> writes:
 >
 >>      y
 >>
 >> x   1    2     4
 >>    1 100 66.67  0
 >>    3  0   0     80
 >>    4  0   33.33 20
 >>
 >> Pointers appreciated.
 >
 >prop.table(hmm,2) * 100
 >

Would it be possible to put a cross reference to this function (and 
margin.table) into the help file for table()?  They are handy functions 
whose existence isn't obvious.

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

After the good observations on "prop.table" as an answer
to Paul Johnson's first problem, and comparing it to my
much less elegant solution....

----- Forwarded message from Jason Turner <jasont at
indigoindustrial.co.nz> -----
Date: Tue, 12 Feb 2002 22:38:12 +1300
From: Jason Turner <jasont at indigoindustrial.co.nz>
To: Paul Johnson <pauljohn at ku.edu>
Subject: Re: [R] A couple of little R things I can't figure out (column
percents, regression with lagged variables)

...
apply(hmm,2,function(zz){zz/sum(zz)}) * 100
...
----- End forwarded message -----

It's my new invention.  I call it "The Wheel".
Off to the patent office ;)

Jason
who can see that the existing wheel is much better than
his, and thinks R is very, very good.  Take a bow,
R-core.
-- 
Indigo Industrial Controls Ltd.
64-21-343-545
jasont at indigoindustrial.co.nz
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._