R help - Apr 2005 - problems with subset (misunderstanding somewhere)

Dear r-help,

I have the following function defined:

cubic.distance<-function(x1,y1,z1,x2,y2,z2) {
  max(c(abs(x1-x2),abs(y1-y2),abs(z1-z2)))
}

I have a data frame from which I make subsets.

When I call
  subset(dataframe,cubic.distance(tb19h,tb37v,tb19v,190,210,227)<=2)
I have the result with 0 rows.

However, the data frame contains the row (among others, that suit)
tb19v tb19h tb37v
226.6 189.3 208.4

Call
   cubic.distance(189.3,208.4,226.6,190,210,227)
gives
[1] 1.6

Next call:
> cubic.distance(189.3,208.4,226.6,190,210,227)<=2
[1] TRUE

It seems to me, that I have made errors somewhere in calls.
Could you, please, be so kind, to tell me, where they are?
Thank you.

--
Best regards
Wladimir Eremeev                                     mailto:wl at eimb.ru

=========================================================================Research
Scientist, PhD                           Leninsky Prospect 33,
Space Monitoring & Ecoinformation Systems Sector, Moscow, Russia, 119071,
Institute of Ecology,                             Phone: (095) 135-9972;
Russian Academy of Sciences                       Fax: (095) 135-9972

On Tue, 5 Apr 2005, Wladimir Eremeev wrote:
> Dear r-help,
>
> I have the following function defined:
>
> cubic.distance<-function(x1,y1,z1,x2,y2,z2) {
>  max(c(abs(x1-x2),abs(y1-y2),abs(z1-z2)))
> }
>
> I have a data frame from which I make subsets.
>
> When I call
>  subset(dataframe,cubic.distance(tb19h,tb37v,tb19v,190,210,227)<=2)
> I have the result with 0 rows.
>
> However, the data frame contains the row (among others, that suit)
> tb19v tb19h tb37v
> 226.6 189.3 208.4
>
> Call
>   cubic.distance(189.3,208.4,226.6,190,210,227)
> gives
> [1] 1.6
>
> Next call:
>> cubic.distance(189.3,208.4,226.6,190,210,227)<=2
> [1] TRUE
>
> It seems to me, that I have made errors somewhere in calls.
> Could you, please, be so kind, to tell me, where they are?
Your function finds the maximum distance over all rows (it is passed 
vectors).  Replace max() by pmax() for a logical result for each row.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

On 05-Apr-05 Wladimir Eremeev wrote:
> Dear r-help,
> 
> I have the following function defined:
> 
> cubic.distance<-function(x1,y1,z1,x2,y2,z2) {
>   max(c(abs(x1-x2),abs(y1-y2),abs(z1-z2)))
> }
> 
> I have a data frame from which I make subsets.
> 
> When I call
>   subset(dataframe,cubic.distance(tb19h,tb37v,tb19v,190,210,227)<=2)
> I have the result with 0 rows.
> 
> However, the data frame contains the row (among others, that suit)
> tb19v tb19h tb37v
> 226.6 189.3 208.4
Did you test the function cubic.distance? As written, I think it
will always return a single value, since max() returns the maximum
of *all* the values, not by rows (even if you use cbind() rather
than c()).

If yor redefine the function as

cubic.distance<-function(x1,y1,z1,x2,y2,z2) {
  apply(cbind(abs(x1-x2),abs(y1-y2),abs(z1-z2)),1,max)
}

I think you will find it does what you want (if I have
understood your problem correctly).

Example (with the function defined as above):

 x<-cbind(rnorm(10,190,1),rnorm(10,210,1),rnorm(10,227,1))
 x<-cbind(rnorm(10,190,2),rnorm(10,210,2),rnorm(10,227,2))
 colnames(x)<-c("tb19h","tb37v","tb19v")
 x.df<-as.data.frame(x)
 (cubic.distance(x.df$tb19h,x.df$tb37v,x.df$tb19v,190,210,227)<=2)
#[1] FALSE FALSE  TRUE  TRUE FALSE FALSE FALSE FALSE  TRUE  TRUE
 subset(x.df,cubic.distance(tb19h,tb37v,tb19v,190,210,227)<=2)
#      tb19h    tb37v    tb19v
#3  189.3930 211.4345 226.3436
#4  189.4521 208.8493 228.0324
#9  188.2441 210.4914 226.4521
#10 191.4781 211.5234 226.1837

With your definition, you would have got the *single"
result FALSE, since there is at least one case where
the distance > 2, so the max > 2, so the subset criterion
evaluates to FALSE, so no rows are selected.

Hoping this helps,
Ted.


--------------------------------------------------------------------
E-Mail: (Ted Harding) <Ted.Harding at nessie.mcc.ac.uk>
Fax-to-email: +44 (0)870 094 0861
Date: 05-Apr-05                                       Time: 14:29:22
------------------------------ XFMail ------------------------------

Ted,

TH> Did you test the function cubic.distance?
Yes, I did.

TH> As written, I think it
TH> will always return a single value,
Yes, here was the misunderstanding.
Subset required a vector, and I gave it a scalar.

Prof. Ripley has already shown my mistake.

--
Best regards
Wladimir Eremeev                                     mailto:wl at eimb.ru

=========================================================================Research
Scientist, PhD                           Leninsky Prospect 33,
Space Monitoring & Ecoinformation Systems Sector, Moscow, Russia, 119071,
Institute of Ecology,                             Phone: (095) 135-9972;
Russian Academy of Sciences                       Fax: (095) 135-9972