On Segmentation of Interaction Values - Developing Concepts in Applied Intelligence - page 30

3. Perform the calculation of the interaction values and establish the interaction
matrix by the submodule INTERACTIONCOMPUTATION.
4. Build and visualize the interaction graph by the submodule INTERAC-
TIONVIEW.

4 Experiments

4.1 Values of Capacity Function

The initial values of capacity function of each interestingness measure is de-
termined by the ratio between the number of different interestingness values
obtained and the total number of association of a rule set. (see table 1).

Table 1. Initially capacity values of interestingness measures.

Interestingness
measures

Initially capacity
values

Interestingness mea-
sures

Initially capacity
values

Conviction 0.54 Confidence 0.002

EII ( α =1) 0.604 Dependency 0.062

EII ( α =2) 0.536 Phi-Coecient 0.014

F-measure 0.016 Rule Interest 0.438

Implication index 0.704 Support 0.086

... ... ... ...

With the initially capacity values in the table 1, we can calculate the capacity
value of Sugeno for each interestingness measure by the method ”Singleton Fuzzy
Measure Ratio Standard” [10] (see table 2). Table 2 describes in detail the values
of the Sugeno's capacity function obtained from the rule set.

Table 2. The capacity value of Sugeno for each interestingness measure.

Interestingness measures Sugeno's
capacity
values

Interestingness measures Sugeno's
capacity
values

Causal Confidence 0.0002 J-measure 0.0384

Causal Confirm 0.0020 Kappa 0.0017

Collective Strength 0.0082 Least Contradiction 0.0037

Confidence 0.0002 Lerman 0.0037

... ... ... ...