R help - Jan 2011 - holding objects in dataframes

Hello list,

I need to keep track of objects that are related to particular
observations.  In this case, I need to keep track of polygons that are
associated with observations.  What I would ideally have is one column
of a dataframe hold a "polygonal" object (from the spatstat package). 
My question: I seem to have managed to do it, but as I haven't read that
dataframes are supposed to be able to hold list elements or objects this
way, I wanted to ask if I'm violating things here in a way such that my
code may break in future R releases.  (that is, am i relying on
non-guaranteed behavior).  My other option is to created a named list or
something along those lines, and just match() the dataframe and named
list when i need access to the related objects.  Please see below for
the code.  Thoughts and suggestions are greatly appreciated!

By tinkering around, I've found that dataframe columns do seem to be
able to hold list elements:

a=data.frame(i=c(1,2),j=c(3,4))
z=list(x=c(1,2,3))
a$k=NA
a[1,]$k=z
> str(a)
    'data.frame':   2 obs. of  3 variables:
     $ i: num  1 2
     $ j: num  3 4
     $ k:List of 2
      ..$ : num  1 2 3
      ..$ : logi NA


Extending that behavior, I can attach "polygonal" objects to
dataframes
it seems:

library(spatstat)
a=data.frame( a=c(1,2), b=c(3,4) )
a$polygons=NA
a[1,]$polygons = list(disc(radius=50))
> str(a)
    'data.frame':   2 obs. of  3 variables:
     $ a       : num  1 2
     $ b       : num  3 4
     $ polygons:List of 2
      ..$ :List of 5
      .. ..$ type  : chr "polygonal"
      .. ..$ xrange: num  -50 50
      .. ..$ yrange: num  -50 50
      .. ..$ bdry  :List of 1
      .. .. ..$ :List of 4
      .. .. .. ..$ x   : num  50 49.9 49.8 49.5 49 ...
      .. .. .. ..$ y   : num  0 2.45 4.9 7.34 9.75 ...
      .. .. .. ..$ area: num 7851
      .. .. .. ..$ hole: logi FALSE
      .. ..$ units :List of 3
      .. .. ..$ singular  : chr "unit"
      .. .. ..$ plural    : chr "units"
      .. .. ..$ multiplier: num 1
      .. .. ..- attr(*, "class")= chr "units"
      .. ..- attr(*, "class")= chr "owin"
      ..$ : logi NA
> area.owin(a[1,]$polygons[[1]])
[1] 7850.828

thanks,
allie

I haven't seen an answer so far, hence I try:

As far as I can see, everything is correct here, since you have an 
element polygons in you data.frame that is a valid vector of mode list.

Nevertheless, since printing /plotting etc. on the dataframe won't give 
desired results now, it would be more natural to introduce a new class 
for your kind of object and write some nice methods for it.

Best,
Uwe Ligges





On 14.01.2011 21:59, Alexander Shenkin wrote:
> Hello list,
>
> I need to keep track of objects that are related to particular
> observations.  In this case, I need to keep track of polygons that are
> associated with observations.  What I would ideally have is one column
> of a dataframe hold a "polygonal" object (from the spatstat
package).
> My question: I seem to have managed to do it, but as I haven't read
that
> dataframes are supposed to be able to hold list elements or objects this
> way, I wanted to ask if I'm violating things here in a way such that my
> code may break in future R releases.  (that is, am i relying on
> non-guaranteed behavior).  My other option is to created a named list or
> something along those lines, and just match() the dataframe and named
> list when i need access to the related objects.  Please see below for
> the code.  Thoughts and suggestions are greatly appreciated!
>
> By tinkering around, I've found that dataframe columns do seem to be
> able to hold list elements:
>
> a=data.frame(i=c(1,2),j=c(3,4))
> z=list(x=c(1,2,3))
> a$k=NA
> a[1,]$k=z
>> str(a)
>
>      'data.frame':   2 obs. of  3 variables:
>       $ i: num  1 2
>       $ j: num  3 4
>       $ k:List of 2
>        ..$ : num  1 2 3
>        ..$ : logi NA
>
>
> Extending that behavior, I can attach "polygonal" objects to
dataframes
> it seems:
>
> library(spatstat)
> a=data.frame( a=c(1,2), b=c(3,4) )
> a$polygons=NA
> 	
>> str(a)
>
>      'data.frame':   2 obs. of  3 variables:
>       $ a       : num  1 2
>       $ b       : num  3 4
>       $ polygons:List of 2
>        ..$ :List of 5
>        .. ..$ type  : chr "polygonal"
>        .. ..$ xrange: num  -50 50
>        .. ..$ yrange: num  -50 50
>        .. ..$ bdry  :List of 1
>        .. .. ..$ :List of 4
>        .. .. .. ..$ x   : num  50 49.9 49.8 49.5 49 ...
>        .. .. .. ..$ y   : num  0 2.45 4.9 7.34 9.75 ...
>        .. .. .. ..$ area: num 7851
>        .. .. .. ..$ hole: logi FALSE
>        .. ..$ units :List of 3
>        .. .. ..$ singular  : chr "unit"
>        .. .. ..$ plural    : chr "units"
>        .. .. ..$ multiplier: num 1
>        .. .. ..- attr(*, "class")= chr "units"
>        .. ..- attr(*, "class")= chr "owin"
>        ..$ : logi NA
>
>> area.owin(a[1,]$polygons[[1]])
> [1] 7850.828
>
> thanks,
> allie
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.