You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by Stefan Kreuzer <st...@aol.de> on 2013/02/04 19:56:59 UTC

ClusterOutputPostProcessorDriver - strange numbering of generated output foldersas

Two questions:
a) Sometimes the ClusterOutputPostProcessorDriver leaves gaps in the 
number of the generated folder, e.g. after folder 15 folder 16 is 
generated. The total number of folders is right (as much folders as 
there were clusters produced in the previous step. Ist this is a bug? 
It causes me some concern, as I (naively) planned to iterate over the 
folder with a simple counter variable.

b) What is the right way to detect what 
ClusterOutputPostProcessorDriver  output folder corresponds to what 
cluster? I try to construct a hierarchical structure, so it is vital 
that the clusters produced based on the 
ClusterOutputPostProcessorDriver output get assigned to their "legal" 
father.

Re: ClusterOutputPostProcessorDriver - strange numbering of generated output foldersas

Posted by Stefan Kreuzer <st...@aol.de>.
Sorry,  sure, I meant 17 instead of 16... too tired. Thanks for the 
info, so there's nothing to worry about.


-----Ursprüngliche Mitteilung-----
Von: Jeff Eastman <jd...@windwardsolutions.com>
An: user <us...@mahout.apache.org>
Verschickt: Di, 5 Feb 2013 12:03 am
Betreff: Re: ClusterOutputPostProcessorDriver - strange numbering of 
generated output foldersas


Maybe a typo? I would expect folder 16 to follow folder 15. For many
reasons though, the cluster numbers may not be monotonic. Suggest you
just iterate over the directories that are presented, their names 
should
correspond to the clusterIds that exist in you clusters-final directory.

On 2/4/13 1:56 PM, Stefan Kreuzer wrote:
> Two questions:
> a) Sometimes the ClusterOutputPostProcessorDriver leaves gaps in the
> number of the generated folder, e.g. after folder 15 folder 16 is
> generated. The total number of folders is right (as much folders as
> there were clusters produced in the previous step. Ist this is a bug?
> It causes me some concern, as I (naively) planned to iterate over the
> folder with a simple counter variable.
>
> b) What is the right way to detect what
> ClusterOutputPostProcessorDriver  output folder corresponds to what
> cluster? I try to construct a hierarchical structure, so it is vital
> that the clusters produced based on the
> ClusterOutputPostProcessorDriver output get assigned to their "legal"
> father.
>
>


  

Re: ClusterOutputPostProcessorDriver - strange numbering of generated output foldersas

Posted by Jeff Eastman <jd...@windwardsolutions.com>.
Maybe a typo? I would expect folder 16 to follow folder 15. For many 
reasons though, the cluster numbers may not be monotonic. Suggest you 
just iterate over the directories that are presented, their names should 
correspond to the clusterIds that exist in you clusters-final directory.

On 2/4/13 1:56 PM, Stefan Kreuzer wrote:
> Two questions:
> a) Sometimes the ClusterOutputPostProcessorDriver leaves gaps in the 
> number of the generated folder, e.g. after folder 15 folder 16 is 
> generated. The total number of folders is right (as much folders as 
> there were clusters produced in the previous step. Ist this is a bug? 
> It causes me some concern, as I (naively) planned to iterate over the 
> folder with a simple counter variable.
>
> b) What is the right way to detect what 
> ClusterOutputPostProcessorDriver  output folder corresponds to what 
> cluster? I try to construct a hierarchical structure, so it is vital 
> that the clusters produced based on the 
> ClusterOutputPostProcessorDriver output get assigned to their "legal" 
> father.
>
>