R help - Nov 2004 - adjusting the map of France to 1830

I'm doing some analyses of historical data from France in 1830 on 'moral
statistics' that I'd like to
show on a map.  I've done most of my analyses in SAS, but a few things 
would work better in R.
To do this, I have to adjust the modern map,

library(maps)
map('france')

to adjust for changes in departments (86 in 1830, to 97 now).  I've read 
the documentation
for the maps and maptools package, but there seems to be no functions to 
allow this, and
I can't find information on the exact structure of map datasets, but I 
understand them to
be delimited lists of polygon coordinates.

In SAS, all maps have (one or more) ID variables representing the 
geographical region,
and there is also a proc gremove that can remove internal boundaries 
inside the polygons
for regions with the same ID.  Is there some way I can do this in R?

Here's what I did in SAS:

*-- Fix the map of France to conform to Guerry:
    - adjust the 97 current departments to correspond to the 86 in 1830
    - delete those not part of France then
;

data gfrtemp;
    set maps.france;
    /* Corse was one dept - merge these to one area, new ID */
    if id in (201, 202)    then dept=200;

    /* Seine et Oise (78) was cut into
    Essonne (91), Val d'Oise (95) and Yvelines (78) */
    else if id in (91, 95)    then dept=78;

    /* Seine (75) now split into
    Hauts-de-Seine (92), Seine-Saint-Denis (93) et Val-de-Marne (94)*/
    else if id in (92, 93, 94)    then dept=75;
   
     /* departments not part of France in 1830 */
    else if id in (
        6,     /* Alpes-Maritimes */
        73,74, /* Savoie, Haute-Savoie */
        90)    /* Territore-de-Belfort */
        then delete;
    else                       dept=id;
    run;
   
*-- remove internal boundaries based on merged DEPT;
proc sort data=gfrtemp;
    by dept;

proc gremove data=gfrtemp out=gfrance;
   by dept;
   id id;
   run;



-- 
Michael Friendly     Email: friendly at yorku.ca 
Professor, Psychology Dept.
York University      Voice: 416 736-5115 x66249 Fax: 416 736-5814
4700 Keele Street    http://www.math.yorku.ca/SCS/friendly.html
Toronto, ONT  M3J 1P3 CANADA

Hello. I do not know if you can merge polygons, but you can select easily:

 > departements=map('france',namesonly=T) # returns a vector of names
of
regions
 > map('france',regions=departements[1:20],namesonly=T) # use what
you need
with regions argument

Hope this helps,


At 16:29 18/11/2004, Michael Friendly wrote:
>I'm doing some analyses of historical data from France in 1830 on
'moral
>statistics' that I'd like to
>show on a map.  I've done most of my analyses in SAS, but a few things 
>would work better in R.
>To do this, I have to adjust the modern map,
>
>library(maps)
>map('france')
>
>to adjust for changes in departments (86 in 1830, to 97 now).  I've read
>the documentation
>for the maps and maptools package, but there seems to be no functions to 
>allow this, and
>I can't find information on the exact structure of map datasets, but I 
>understand them to
>be delimited lists of polygon coordinates.
>
>In SAS, all maps have (one or more) ID variables representing the 
>geographical region,
>and there is also a proc gremove that can remove internal boundaries 
>inside the polygons
>for regions with the same ID.  Is there some way I can do this in R?
>
>Here's what I did in SAS:
>
>*-- Fix the map of France to conform to Guerry:
>    - adjust the 97 current departments to correspond to the 86 in 1830
>    - delete those not part of France then
>;
>
>data gfrtemp;
>    set maps.france;
>    /* Corse was one dept - merge these to one area, new ID */
>    if id in (201, 202)    then dept=200;
>
>    /* Seine et Oise (78) was cut into
>    Essonne (91), Val d'Oise (95) and Yvelines (78) */
>    else if id in (91, 95)    then dept=78;
>
>    /* Seine (75) now split into
>    Hauts-de-Seine (92), Seine-Saint-Denis (93) et Val-de-Marne (94)*/
>    else if id in (92, 93, 94)    then dept=75;
>
>     /* departments not part of France in 1830 */
>    else if id in (
>        6,     /* Alpes-Maritimes */
>        73,74, /* Savoie, Haute-Savoie */
>        90)    /* Territore-de-Belfort */
>        then delete;
>    else                       dept=id;
>    run;
>
>*-- remove internal boundaries based on merged DEPT;
>proc sort data=gfrtemp;
>    by dept;
>
>proc gremove data=gfrtemp out=gfrance;
>   by dept;
>   id id;
>   run;
>
>
>
>--
>Michael Friendly     Email: friendly at yorku.ca Professor, Psychology Dept.
>York University      Voice: 416 736-5115 x66249 Fax: 416 736-5814
>4700 Keele Street    http://www.math.yorku.ca/SCS/friendly.html
>Toronto, ONT  M3J 1P3 CANADA
>
>______________________________________________
>R-help at stat.math.ethz.ch mailing list
>https://stat.ethz.ch/mailman/listinfo/r-help
>PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html
St??phane DRAY
--------------------------------------------------------------------------------------------------

D??partement des Sciences Biologiques
Universit?? de Montr??al, C.P. 6128, succursale centre-ville
Montr??al, Qu??bec H3C 3J7, Canada

Tel : (514) 343-6111 poste 1233         Fax : (514) 343-2293
E-mail : stephane.dray at umontreal.ca
--------------------------------------------------------------------------------------------------

Web                                          http://www.steph280.freesurf.fr/

At 16:29 18/11/2004, Michael Friendly wrote:
> I'm doing some analyses of historical data from France in 1830 on
'moral
> statistics' that I'd like to
> show on a map.  I've done most of my analyses in SAS, but a few things 
> would work better in R.
> To do this, I have to adjust the modern map,
> 
> library(maps)
> map('france')
> 
> to adjust for changes in departments (86 in 1830, to 97 now).  I've
read
> the documentation
> for the maps and maptools package, but there seems to be no functions to 
> allow this, and
> I can't find information on the exact structure of map datasets, but I 
> understand them to
> be delimited lists of polygon coordinates.
> 
> In SAS, all maps have (one or more) ID variables representing the 
> geographical region,
> and there is also a proc gremove that can remove internal boundaries 
> inside the polygons
> for regions with the same ID.  Is there some way I can do this in R?
>
Unfortunately not with the current implementation of several of the
'extra' databases in the mapdata package.  The map() function does have
the interior=FALSE option, which would normally do what you want, but
only when the data has been manipulated to allow it.  Currently this
option is only useful with the world and usa maps (and their
derivatives, such as world2 and state).

Currently every department is a complete polygon, and so every interior
line segment occurs twice in the data.  What has to happen to the data
is for it to be split up into non-overlapping line segments, and each
polygon reconstructed from a list of these line segments (with
direction being important).

If you are prepared to perform this somewhat tedious process, I am
happy to assist you with further details.

However even with the interior= option functioning, it would still not
be easy to produce the map you would require, since you would have to
build it up from many components (namely each of the 'combined'
departments, plus 'all the rest').

HTH,
Ray Brownrigg

Nathan Leon Pace, MD, MStat wrote:
> Hi,
> 
> Using R 2.0.1 on Mac g5 running Mac OS X 10.3.6.
> 
> I would expect that
> 
> abs(.7 - .5) >= abs(.3 - .5) should be returned TRUE.
> 
> Instead
> 
>  > www <- abs(.7 - .5) >= abs(.3 - .5)
>  > www
> [1] FALSE
> 
> Is this a result of floating point or the implementation of abs or 
> something else?
Due to floating point arithmetic in general, not specifically the abs 
function.  The number .5 will have an exact floating point 
representation but .3 and .7 will not.  Making equality comparisons with 
floating point values is always risky.
> In a function I need to compare two absolute values - each being of the 
> form |variable - constant|.
On a Mac I get

 > abs(.7-.5) - abs(.3-.5)
[1] -5.551115e-17

so you need to make a relative comparison, not an absolute comparison. 
You could write the relative comparison using the all.equal function, 
such as

 > v1 <- abs(.7-.5)
 > v2 <- abs(.3-.5)
 > (v1 > v2) || all.equal(v1, v2)
[1] TRUE

> Date: Fri, 19 Nov 2004 15:59:25 -0500
> From: Michael Friendly <friendly at yorku.ca>
> 
> Here's what I tried.   I can plot a selection of regions, but I
> can't seem to remove an arbitrary list of region numbers, unless
I've
> done something wrong
> by selecting the regions I want to plot with departements[-exclude].
I think here the problemis not using exact=T in the call to map(), see
below.
> I also get an error
> when I try to use map.text to label a map with only the regions I'm  
> selecting.
> 
>  > departements <- map('france',namesonly=T, plot=FALSE)
>  > # returns a vector of names of regions
>  >
>  > exclude <- c(47,  #Alpes-Maritimes
> + 66,  # Haute-Savoie
> + 76,  # Savoie
> + 95,  # Territore-de-Belfort
> + 109, 110, 111, # Var: Iles d'Hyeres
> + 49, 53, 54, 55, # Moribhan: Isles
> + 62, 64,    # Vendee: Isles
> + 72, 75     # Charente-Maritime: Isles
> + )
>  >
>  > depts <- departements[-exclude]
>  > gfrance <-map('france', regions=depts)
>  > labels <- (as.character(1:length(departements)))[-exclude]
>  > gfrance <-map.text('france', regions=depts, add=FALSE,
labels=labels)
> Error in map.text("france", regions = depts, add = FALSE, labels
= labels) :
>         map object must have polygons (fill=TRUE)
> 
That error message is issued when regions= specifies less than the whole
database, in which case the alternate "list of 'x', 'y',
and 'names'
obtained from a previous call to 'map'" is required as a first
parameter.

So to do what you want (notwithstanding the next problem you raise), try
the following as a demonstration:

gfrance <- map('france', regions=depts, exact=T, fill=T, plot=F)
map.text(gfrance, regions=depts, labels=labels)  
map('france', regions=departements[exclude], fill=T, col=1, add=T)
> Another problem, potentially more difficult for mapping data on the map 
> of France is that
> the "departements" are actually just the polygons in the map, 
> arbitrarily numbered from
> east to west, and from north to south --- they don't correspond to the 
> 'official' administrative
> region numbers.  As well, the departement names don't always match 
> exactly (ignoring
> accents, e.g., Val-d'Oise vs. Val-Doise) so it would be another 
> challenge to plot my
> historical data on the map of France.
> 
Well, maps is a source package! [:-)].

You are most welcome to modify the source files
maps/src/france.{gon,line,name} to reorder the polygons (and correct
errors in the names).  If the relationship between those 3 files is not
obvious, contact me for further details.  Also, I am happy to fold your
changes back into the original maps package.

Ray Brownrigg