Fuzzy rule learning output


Does anyone know what is the meaning of “weight”, “spread”, “features” and “variance” ?

Is there any detailed explanation available?

2022-07-05 07_49_52-Sorted Table - 3_1663_1724_1722 - Sorter


Hi @yupliu

As mentioned in the -Fuzzy Rule Learner- node, the algorithm implemented in this node is explained in a published paper by @berthold. The paper is available from the konstanz University through the following link:

The terms “Weight” & “Spread” are refereed in a 2nd paper by @berthold & @gab1one also cited in the -Fuzzy Rule Learner- node description and available from here too:

Hope it helps.



Hi Ael,

I quickly scanned these two paper before posting the question here. It seems there is no much information regarding to these four outputs. I have some understanding about “weight” but not others.

Do you know what “Features” mean here? It’s not the total number of input features. Does it mean the number of features used in a rule? If so, it’s not that easy to see which feature is used since every feature has a range in the output table.


Hi @yupliu

The authors (@berthold & @gab1one) of this fuzzy method are mentioned in my previous post and hence they may provide answers to your questions. Besides these two previously cited papers, the following one may also help to understand their underlying fuzzy algorithm:

Fuzzy Logic in KNIME – Modules for Approximate Reasoning –

Their other papers on ResearchGate may be of help too as complementary information.


1 Like

Excellent choice of node :slight_smile:
As for your questions:

  • Weight is the number of training patterns that are ultimately covered (“explained”) by the rule
  • Spread is the size of the core region of the rule
  • Number of features is the number of features that are actually constrained - the rules themselves use the domain information instead of+/- infinity, though, so from those representations it’s a bit harder to see directly (the borders should be min/max resp.)
  • Variance is the size of support region of the rule.

Does this help?

Cheers, Michael


Hi Michael,

This is very helpful, thanks a lot!

1 Like