Altair RISE

A program to recognize and reward our most engaged community members

Nominate Yourself Now!

"comparing clustering algorithms"

bdeshmukh2000

Hi
i want to compare different classification algorithms in rapid miner
i want to evaluate these algorithm based on the parameters like FP.TP, Precision, Rcall, ROC, confusion matrix ---- they way it happens in WEKA...

Can you help me on this ?

Bharat

Find more posts tagged with

AI Studio

Clustering

Accepted answers

All comments

IngoRM

Hi Bharat,

you probably want to compare classification algorithms instead of clustering algorithms if you want to see those measures you stated in your post. Nevertheless, this is of course possible with RapidMiner. Did you already read our manual? Or had a look into our video tutorials? And the sample processes delivered with RapidMiner? Or those available on myExperiment? All necessary information is available on our web site or directly within RapidMiner. If you let us know what problem you have in particular maybe we could better help.

Cheers,
Ingo

bdeshmukh2000

Hi
thanx for the reply.
i am try to compare clustering algorithms using arff data sets . i have read the manual also but not able to figure out the performance evaluation for clustering. i have tried it with evaluation -> performance measurement -> performance and also clustering evaluation. but getting the error for setting up the role.

i have tried with the samples repository also but not able to locate the example .

Thank you......waiting for help.

Bharat

IngoRM

And how do you calculate

parameters like FP.TP, Precision, Rcall, ROC, confusion matrix

? I mean: in theory? Without having any class labels (I assume you don't have labels since why should you want to cluster otherwise)?

Sorry without any more details about your data the only thing left to me is the thing I already did: recommend the videos and sample processes to you. All answers can be found there or here in the forum (searching for example for "cluster evaluation" delivers a lot of hits...)

Cheers,
Ingo