Cluster Interaction

brunonbrasil

hi!
I have tasks in certain clusters, these tasks interact with other tasks, thus forming interaction between clusters. Example: cluster A and B form an interaction or cluster B, C and D form another interaction. Is there any method in the rapidminer that forms groups or new cluster clusters initially identified, and is there any method of evaluating the interaction of clusters, that is, the precision with confusion matrix?

lionelderkrikor

Hi @brunonbrasil,

Mmmhh, the confusion matrix is a performance indicator dedicated to supervised learning : It means that your training set contains a known label (target).
Clustering belongs to unsupervised learning, thus there is no known label in your training set and thus a confusion matrix can not be calculated in such case.
You could maybe elaborate what you are trying to do and explain explicitly the notion called "interaction of clusters".

Regards,

Lionel 

brunonbrasil

Hi @Leonelderkrikor.

I already used matrix confusion for clustering, I have a label on the predefined clusters for each sentence and then I ran in the rapidminer with the matrix confusion and got the desired results.

Now my case is, I have two people who communicate with each other, when using the rapidminer I wanted to form new clusters that would form a cluster with several predefined clusters. For example, Carla interacts with Mariana through clusters c2, m8 and c3 and m1. Knowing that I have this predefined interaction I wanted to use the confusion matrix to evaluate the interaction in the rapidminer.