how to implement distribution for instance for the test set

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

how to implement distribution for instance for the test set

Bill McLaughlin
To whom may help:

I would like to obtain the cluster probabilities per instance for a
test set using a model created with the EM clusterer.  I have used the
following scheme:

weka.clusterers.EM -l model_file -T test_data -p 0

and obtained the output:

"instance"  "most likely cluster number"
.
.
.
etc. for each instance

Please let me know if there is an option to obtain the cluster
probabilities for each test instance when using the cluster model.
The output could be in a format similar to that obtained for
the training instances when using the verbose option.  For example,
using the verbose option part of the output in for the training set is:

"instance" "most likely cluster number"  "cluster probabilty for being
in cluster n" "cluster probabilty of being in cluster n+1" ...etc for
each cluster

Regards,

Bill McLaughlin

_______________________________________________
Wekalist mailing list
[hidden email]
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist