Orange Forum • View topic - Interpreting an interaction graph

Interpreting an interaction graph

A place to ask questions about methods in Orange and how they are used and other general support.

Interpreting an interaction graph

Postby Makandal » Thu Nov 19, 2009 17:46

Sorry, I am not a statistician nor mathematician but a simple basic geologist trying to make sense of too many data (discrete and continuous). So please be patient with me.

I am trying to understand how to interprete the interaction graph vs a hierarchical clustering from an attribute distance matrix.
For example, the presence of mica and the sorting of a sandstone seems to be correlated in HC. Which seems logic in the geological context...

In the interaction graph, sorting is not even represented and I read in the bar graph : mica 48.5%, Sorting 49.7, interaction -23%. What does this mean ?

Also, which class should I chose ? A class I am looking for information (for example, the presence or not of mica or glauconia) or the name of the facies ?

Sorry for my beotian question, I am a real beginner in this interesting domain and my knowledge in data mining is limited to interpretation of PCA...

Return to Questions & Support