You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by David G <da...@gmail.com> on 2013/12/12 17:42:47 UTC
Visualizing cluster trough command line
Hello,
I am doing some experiments with clustering but now I want to visualize the
data. Like in
https://cwiki.apache.org/confluence/display/MAHOUT/Visualizing+Sample+Clusters
,
is there a way to run the classes with arguments that accept custom cluster
data ? What is the best way to see cluster data?
The command i am using is: mvn -q exec:java
-Dexec.mainClass=org.apache.mahout.clustering.display.DisplayClustering
Thank you
--
David Graça
davidgraca@gmail.com
david.graca@ist.utl.pt
Re: Visualizing cluster trough command line
Posted by Suneel Marthi <su...@yahoo.com>.
U can use clusterdump to generate GraphML, CSV, Text and JSON outputs.
mahout clusterdump -i <cluster-output>/clusters-0-final -of GRAPH_ML -o xyz.graphml -p <cluster-output>/clusteredPoints
On Friday, December 13, 2013 7:58 AM, David G <da...@gmail.com> wrote:
I find your idea interesting. Can you elaborate on that?
On 13 December 2013 12:36, Gokhan Capan <gk...@gmail.com> wrote:
> I think ClusterDumper can also export cluster results in GraphML format,
> which can be used to visualize clustered points in a graph visualization
> software such as Gephi.
>
> Gokhan
>
>
> On Fri, Dec 13, 2013 at 10:44 AM, David G <da...@gmail.com> wrote:
>
> > Hi,
> > I see what you're saying. For more than 3 dimensions it begins to be
> > complicated to plot the results. In those cases what is the correct way
> to
> > pick the data and make some sense out of it?
> >
> > Thanks
> >
> >
> > On 13 December 2013 09:28, Taner Diler <ta...@gmail.com> wrote:
> >
> > > Hi David,
> > >
> > > I tried to find a solution to visualize, too. In DisplayClustering
> > example,
> > > clustering is running for x,y vectors and easy to visualize. But in
> real
> > > world, we have n vectors. It's not possible to visualize as point in xy
> > > chart, I think.
> > >
> > >
> > > On Thu, Dec 12, 2013 at 6:42 PM, David G <da...@gmail.com> wrote:
> > >
> > > > Hello,
> > > >
> > > > I am doing some experiments with clustering but now I want to
> visualize
> > > the
> > > > data. Like in
> > > >
> > > >
> > >
> >
> https://cwiki.apache.org/confluence/display/MAHOUT/Visualizing+Sample+Clusters
> > > > ,
> > > > is there a way to run the classes with arguments that accept custom
> > > cluster
> > > > data ? What is the best way to see cluster data?
> > > >
> > > > The command i am using is: mvn -q exec:java
> > > >
> -Dexec.mainClass=org.apache.mahout.clustering.display.DisplayClustering
> > > >
> > > > Thank you
> > > >
> > > > --
> > > > David Graça
> > > > davidgraca@gmail.com
> > > > david.graca@ist.utl.pt
> > > >
> > >
> >
> >
> >
> > --
> > David Graça
> > davidgraca@gmail.com
> > david.graca@ist.utl.pt
> >
>
--
David Graça
davidgraca@gmail.com
david.graca@ist.utl.pt
Re: Visualizing cluster trough command line
Posted by David G <da...@gmail.com>.
I find your idea interesting. Can you elaborate on that?
On 13 December 2013 12:36, Gokhan Capan <gk...@gmail.com> wrote:
> I think ClusterDumper can also export cluster results in GraphML format,
> which can be used to visualize clustered points in a graph visualization
> software such as Gephi.
>
> Gokhan
>
>
> On Fri, Dec 13, 2013 at 10:44 AM, David G <da...@gmail.com> wrote:
>
> > Hi,
> > I see what you're saying. For more than 3 dimensions it begins to be
> > complicated to plot the results. In those cases what is the correct way
> to
> > pick the data and make some sense out of it?
> >
> > Thanks
> >
> >
> > On 13 December 2013 09:28, Taner Diler <ta...@gmail.com> wrote:
> >
> > > Hi David,
> > >
> > > I tried to find a solution to visualize, too. In DisplayClustering
> > example,
> > > clustering is running for x,y vectors and easy to visualize. But in
> real
> > > world, we have n vectors. It's not possible to visualize as point in xy
> > > chart, I think.
> > >
> > >
> > > On Thu, Dec 12, 2013 at 6:42 PM, David G <da...@gmail.com> wrote:
> > >
> > > > Hello,
> > > >
> > > > I am doing some experiments with clustering but now I want to
> visualize
> > > the
> > > > data. Like in
> > > >
> > > >
> > >
> >
> https://cwiki.apache.org/confluence/display/MAHOUT/Visualizing+Sample+Clusters
> > > > ,
> > > > is there a way to run the classes with arguments that accept custom
> > > cluster
> > > > data ? What is the best way to see cluster data?
> > > >
> > > > The command i am using is: mvn -q exec:java
> > > >
> -Dexec.mainClass=org.apache.mahout.clustering.display.DisplayClustering
> > > >
> > > > Thank you
> > > >
> > > > --
> > > > David Graça
> > > > davidgraca@gmail.com
> > > > david.graca@ist.utl.pt
> > > >
> > >
> >
> >
> >
> > --
> > David Graça
> > davidgraca@gmail.com
> > david.graca@ist.utl.pt
> >
>
--
David Graça
davidgraca@gmail.com
david.graca@ist.utl.pt
Re: Visualizing cluster trough command line
Posted by Gokhan Capan <gk...@gmail.com>.
I think ClusterDumper can also export cluster results in GraphML format,
which can be used to visualize clustered points in a graph visualization
software such as Gephi.
Gokhan
On Fri, Dec 13, 2013 at 10:44 AM, David G <da...@gmail.com> wrote:
> Hi,
> I see what you're saying. For more than 3 dimensions it begins to be
> complicated to plot the results. In those cases what is the correct way to
> pick the data and make some sense out of it?
>
> Thanks
>
>
> On 13 December 2013 09:28, Taner Diler <ta...@gmail.com> wrote:
>
> > Hi David,
> >
> > I tried to find a solution to visualize, too. In DisplayClustering
> example,
> > clustering is running for x,y vectors and easy to visualize. But in real
> > world, we have n vectors. It's not possible to visualize as point in xy
> > chart, I think.
> >
> >
> > On Thu, Dec 12, 2013 at 6:42 PM, David G <da...@gmail.com> wrote:
> >
> > > Hello,
> > >
> > > I am doing some experiments with clustering but now I want to visualize
> > the
> > > data. Like in
> > >
> > >
> >
> https://cwiki.apache.org/confluence/display/MAHOUT/Visualizing+Sample+Clusters
> > > ,
> > > is there a way to run the classes with arguments that accept custom
> > cluster
> > > data ? What is the best way to see cluster data?
> > >
> > > The command i am using is: mvn -q exec:java
> > > -Dexec.mainClass=org.apache.mahout.clustering.display.DisplayClustering
> > >
> > > Thank you
> > >
> > > --
> > > David Graça
> > > davidgraca@gmail.com
> > > david.graca@ist.utl.pt
> > >
> >
>
>
>
> --
> David Graça
> davidgraca@gmail.com
> david.graca@ist.utl.pt
>
Re: Visualizing cluster trough command line
Posted by David G <da...@gmail.com>.
Hi,
I see what you're saying. For more than 3 dimensions it begins to be
complicated to plot the results. In those cases what is the correct way to
pick the data and make some sense out of it?
Thanks
On 13 December 2013 09:28, Taner Diler <ta...@gmail.com> wrote:
> Hi David,
>
> I tried to find a solution to visualize, too. In DisplayClustering example,
> clustering is running for x,y vectors and easy to visualize. But in real
> world, we have n vectors. It's not possible to visualize as point in xy
> chart, I think.
>
>
> On Thu, Dec 12, 2013 at 6:42 PM, David G <da...@gmail.com> wrote:
>
> > Hello,
> >
> > I am doing some experiments with clustering but now I want to visualize
> the
> > data. Like in
> >
> >
> https://cwiki.apache.org/confluence/display/MAHOUT/Visualizing+Sample+Clusters
> > ,
> > is there a way to run the classes with arguments that accept custom
> cluster
> > data ? What is the best way to see cluster data?
> >
> > The command i am using is: mvn -q exec:java
> > -Dexec.mainClass=org.apache.mahout.clustering.display.DisplayClustering
> >
> > Thank you
> >
> > --
> > David Graça
> > davidgraca@gmail.com
> > david.graca@ist.utl.pt
> >
>
--
David Graça
davidgraca@gmail.com
david.graca@ist.utl.pt
Re: Visualizing cluster trough command line
Posted by Taner Diler <ta...@gmail.com>.
Hi David,
I tried to find a solution to visualize, too. In DisplayClustering example,
clustering is running for x,y vectors and easy to visualize. But in real
world, we have n vectors. It's not possible to visualize as point in xy
chart, I think.
On Thu, Dec 12, 2013 at 6:42 PM, David G <da...@gmail.com> wrote:
> Hello,
>
> I am doing some experiments with clustering but now I want to visualize the
> data. Like in
>
> https://cwiki.apache.org/confluence/display/MAHOUT/Visualizing+Sample+Clusters
> ,
> is there a way to run the classes with arguments that accept custom cluster
> data ? What is the best way to see cluster data?
>
> The command i am using is: mvn -q exec:java
> -Dexec.mainClass=org.apache.mahout.clustering.display.DisplayClustering
>
> Thank you
>
> --
> David Graça
> davidgraca@gmail.com
> david.graca@ist.utl.pt
>