Configuration file

The configuration file (.fs) must follow the following structure:

---

Classification/Regression/Association
C1(100),C2(50), .../Low, Medium,...(350) .../Vble_1(LABEL_0,LABEL_1), Vble_2(LABEL_0), ...

Source: GUAJE/FURIA/KEEL/...
Blank threshold: 0.1
Goodness Threshold (high): 0.5
Goodness Threshold (low): 0.1

Rules: 11
Examples: 102

Rule1: If Vble_1 is Low THEN Class is C1
Correct Conclusion => (22.93300) (0.63703) => 98(0.333), 99(0.264), 100(0.123), ...
Incorrect Conclusion => (4.89900) (0.22268) => 17(0.109), 21(0.235), 47(0.212), ...
Positive Examples => (20.47700) (0.68257) => 98(0.333), 100(0.123), 102(0.421), ...
Negative Examples => (4.89900) (0.22268) => 17(0.109), 21(0.235), 47(0.212), ...

Rule2: If Vble_2 is Medium THEN Vble_Output is Low
Correct Conclusion => (3.00300) (0.25025) => (1.0), 14(0.352), 17(0.829), ...
Incorrect Conclusion => (0.00000) (0.00000) => There are no items
Positive Examples => (0.00000) (0.00000) => There are no items
Negative Examples => (0.00000) (0.00000) => There are no items

Rule3: IF Vble_3 is LABEL_1 THEN Vble_4 is LABEL_0
Rule Support => (737.32151) (0.67830) => 0(0.561), 1(0.407), 2(0.872), ...
Antecedent Support => (747.00990) (0.68722) => 0(0.561), 1(0.407), 2(0.872), ...
Consequent Support => (998.73089) (0.91879) => 0(0.561), 1(0.407), 2(0.872), ...

...

Rule11: UNCOVERED INSTANCES
80 (C1), 90(C1), 93(C2), ... / 1, 2, 3, ... / 4, 5, 6

---

• The first line indicates the type of fingram you wanna create: Classification, Regression or Association.
• Second line changes depending on the selected type of rules. It includes the classes (with the number of examples per class in brackets) for classification problems; the output linguistic labels and the total number of examples at the end of the line (in brackets) in case of regression problems; and the possible output variables along with the attached linguistic terms (in brackets) for association rules.
• The next four lines show:
  • The software used to create the .fs file.
  • Blank threshold represents the minimum level an example should fire a rule to be taken into account.
  • Goodness thresholds (low and high) correspond to the minimum and maximum levels of rule firing that are considered when computing the goodness measure [D. P. Pancho, J. M. Alonso, O. Cordón A. Quirin, and L. Magdalena, "FINGRAMS: visual representations of fuzzy rule-based inference for expert analysis of comprehensibility", IEEE Trans. on Fuzzy Systems, doi: 10.1109/TFUZZ.2013.2245130, 2013.][9] in classification and regression problems.
• Next line yields the total number of rules.
• The following lines detail information related to each rule. For illustrative purposes, we have included three examples of rules (each one corresponding to one of the three types of rules that are considered). Thus, Rule1, Rule2, and Rule3 exemplify rules of classification, regression and association respectively. In all cases the first line gives the number of the rule along with its linguistic description.
• Last two lines show how to specify uncovered instances in this configuration file. First line shows the text UNCOVERED INSTANCES. Second line includes the identifier of instances not covered by the set of rules. Only in case of classification rules this identifier is accompanied by the output clas of the instance in brackets.