How to overcome limitations of k-means related to Categorical Data

Hello ,

I am new to knime community , i want to know is there any way to use categorical data in K mean?

and one more can i draw elbow curve in knime? how can i calculate purity of cluster?

 

 

Hi SGK,

There are two different approaches here. You can either use k-medoids instead of k-means, or you can use k-means, but then you need to create dummy variables for the catagories first using the One to Many node.

For using the elbow method, please have a look at the example on our Example Server in 08_Other_Analytics_Types/01_Text_Processing/17_Topic_Extraction_with_the_Elbow_Method

Hope that helps!

Cheers,

Roland

thanks ronald

i cant exactly locate this file..

i want to know number of clusters using elbow curve

regards

 

thanks ronald

i cant exactly locate this file..

i want to know number of clusters using elbow curve

regards