"Using SOM clustering o/p to re-create original table with cluster-id column"

Question

I have performed the SOM based Clustering using the following operators:
 - DatabaseExampleSource
 - SOMDimensionalityReduction
 - ChangeAttributeRole

The i/p table format is here:
id attr1  attr2 ....... attr50
-------------------------------
0 val1   val2..................       
1 val3   val4..................
.........

I get the output like this:
id SOM_0
-----------
0     2
1     0
....

I would like to create a Table in the Database to include the cluster column with the original table. The new table should look like this: 
id attr1  attr2 ....... attr50    SOM_0
------------------------------------------
0 val1   val2..................        2
1 val3   val4..................        0
.........

How do I achieve it using the Rapidminer operators?

TobiasMalbrecht · Answer

Hi,ruser  wrote:
I used the DatabaseExampleSetWriter to write the o/p of the SOM clustering into a new DB Table. Then, I can perform the Table 'join' operation to combine the original table with this new table. In the end, I have the desired result with the original table re-created with the Cluster-id column.

You can do the same using RapidMiner by using the [tt]ExampleSetJoin[/tt] operator. You might also need to use the [tt]IOMultiplier[/tt] to create a copy of the original example set before you apply the clustering algorithm.ruser  wrote:
I like working with the Rapidminer.

Thanks, that's great.

Kind regards,
Tobias

ruser · Answer

Great! I found the answer myself :-)

I used the DatabaseExampleSetWriter to write the o/p of the SOM clustering into a new DB Table. Then, I can perform the Table 'join' operation to combine the original table with this new table. In the end, I have the desired result with the original table re-created with the Cluster-id column.

I like working with the Rapidminer.