[R-sig-Geo] Package for Classification????

[Robert J. Hijmans]

For supervised classification the advantage of R could be that you
have more a more flexible algorithms available. I know some folks have
been using RandomForest together with 'raster' so that you can do
large files. You can have a look at the 'dismo' package which talks
about species distribution modeling (but the flow of data analysis is
the same for classification).

For unsupervised classification the R algorithms are nice too but they
have data size limitations that cannot be solved by 'raster'.

Robert


On Wed, Nov 17, 2010 at 4:46 AM, Lucy Tallents
<conservationgis at btinternet.com> wrote:
> Dear Malik,
>
>
>
> I'm a newcomer to GRASS and R myself, having recently moved over from
> ArcGIS/Erdas, so I'm probably not the best person to assist you.  My
> suggestion was merely that it seems sensible to use a GIS software to do a
> GIS task, particularly one such as classification which often requires a
> large degree of interaction with the data to define training samples and
> improve the classification iteratively.
>

Or do data analysis with data analysis software (such as R)?

>
>
> I'm interested to hear opinions from others - is GRASS the best route for
> land-cover classification, or are stand-alone packages in R a viable
> alternative?  Any advice would be much appreciated - thank you.
>
>
>
> Regarding the use of GRASS and R, my understanding is that you start R
> through the GRASS terminal and then work in the usual R interface.  This
> presumably allows R to apply GRASS commands to the GIS layers in the
> location & mapset you started GRASS in.  This booting of R through GRASS is
> briefly described at the end of the
> http://grass.fbk.eu/statsgrass/grass6_r_install.html.
>
>
>
> If you need help in installing the xml package (from the list, not from
> newbie me!), you would need to provide more information about your setup
> (operating system/R version), what method you used to install (download zip
> file, install directly through R?), and the precise error message.  However,
> I strongly suggest that you use the online help and maillist archives to try
> and find your answer before sending a request to the list.
>
>
>
> I hope that helps - other readers please do feel free to correct any errors
> or omissions of mine.
>
>
>
> Best wishes,
>
>
>
> Lucy
>
>
>
> From: Malik Shahzad [mailto:geomalik at live.com]
> Sent: 17 November 2010 08:24
> To: conservationgis at btinternet.com
> Subject: RE: [R-sig-Geo] Package for Classification????
>
>
>
> Dear Dr. Lucy!
>
>
>
> Thanks for your kind help. As per your instructions I tried to install
> GRASS. Then I tried to load package "spgrass6" in R and found error:
>
> Need "XML" Package and I am unable to install  XML package for windows...
> from
>
> http://cran.r-project.org/web/packages/XML <http://>
>
>
>
> Second, I will work in R-Interface after loading GRASS package or I have to
> work in GRASS Interface???
>
>
>
> Please help me in this regards, As I am new in both packages... & once again
> thanks...
>
> with Best Regards,
>
> Malik Shahzad
> Visiting Researcher
> National Institute of Informatics (NII)
> Tokyo, Japan
>
> Doctoral Student
> Asian Institute of Technology (AIT)
> Bangkok, Thailand
> +66-8-7676-5616
>
>
>
>
>  _____
>
> From: conservationgis at btinternet.com
> To: geomalik at live.com
> Subject: RE: [R-sig-Geo] Package for Classification????
> Date: Tue, 16 Nov 2010 16:21:36 +0000
>
> Dear Malik,
>
>
>
> Using the GRASS-R interface would seem to be a sensible route.  See:
>
>
>
> http://cran.r-project.org/web/packages/spgrass6/index.html
>
> http://grass.fbk.eu/statsgrass/grass6_r_install.html
>
>
>
> Best wishes,
>
>
>
> Lucy
>
>
>
>
>
> Dr Lucy Tallents
>
> Assistant Coordinator, PG Dip in International Wildlife Conservation
> Practice
>
>
>
> Wildlife Conservation Research Unit (WildCRU)
>
> Zoology, University of Oxford
>
> The Recanati-Kaplan Centre
>
> Tubney House, Abingdon Road
>
> Tubney, Abingdon OX13 5QL, UK
>
> T +44 (0) 1865 611125
>
> F +44 (0) 1865 611101
>
> M +44 (0) 7763 378147
>
>  <http://www.wildcru.org/diploma> www.wildcru.org/diploma
>
>
>
>
>
>
>
> From: r-sig-geo-bounces at stat.math.ethz.ch
> [mailto:r-sig-geo-bounces at stat.math.ethz.ch] On Behalf Of Malik Shahzad
> Sent: 16 November 2010 16:16
> To: R Help
> Subject: [R-sig-Geo] Package for Classification????
>
>
>
>
>
> Dear I want to know a good package for Raster data classification (
> Supervised & Unsupervised).Thanks in Advance...
>
> with Best Regards,
>
>
>
> Malik Shahzad
>
> Visiting Researcher
>
> National Institute of Informatics (NII)
>
> Tokyo, Japan
>
>
>
> Doctoral Student
>
> Asian Institute of Technology (AIT)
>
> Bangkok, Thailand
>
> +66-8-7676-5616
>
>
>
>        [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
>
>
>  _____
>
> No virus found in this message.
> Checked by AVG - www.avg.com
>
>
>
>        [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>