kmeans question
Hello i am new in Rapidminer and i d like to ask something. Is there a way to view results of kmeans clustering in text mode?something like this:
eg. cluster 0 cluster 1
age 39.4 52.3
state NY DC
salary 15132.7 52789.5
I take the mean of each attribute when it is numeric and the most used choice when nominal.Another question is that when i use the kmean (not the weka one) it cant handle any nominal attributes!!!
I dont want the state or sex to become numeric!Then i lose useful data!
eg. cluster 0 cluster 1
age 39.4 52.3
state NY DC
salary 15132.7 52789.5
I take the mean of each attribute when it is numeric and the most used choice when nominal.Another question is that when i use the kmean (not the weka one) it cant handle any nominal attributes!!!
I dont want the state or sex to become numeric!Then i lose useful data!
0
Answers
you can create an example set consisting of the medoids of each cluster with the Extract Cluster Prototypes operator.
To cluster nominal attributes you can either try to use k-Medoids which offers a rich set of nominal distance measures, or you can code your nominal attributes via e.g. dummy coding using the Nominal to Numeric operator and then use k-Means. To use dummy coding, make sure that you are using the latest version of RapidMiner.
Cheers,
Marius
On the results i only take the overall stats.