Tertius

Package

weka.associations

Finds rules according to confirmation measure (Tertius-type algorithm).

For more information see:

P. A. Flach, N. Lachiche (1999). Confirmation-Guided Discovery of first-order rules with Tertius. Machine Learning. 42:61-95.

The table below describes the options available for Tertius.

Option	Description
classIndex	Index of the class attribute. If set to 0, the class will be the last attribute.
classification	Find only rules with the class in the head.
confirmationThreshold	Minimum confirmation of the rules.
confirmationValues	Number of best confirmation values to find.
frequencyThreshold	Minimum proportion of instances satisfying head and body of rules
hornClauses	Find rules with a single conclusion literal only.
missingValues	Set the way to handle missing values. Missing values can be set to match any value, or never match values or to be significant and possibly appear in rules.
negation	Set the type of negation allowed in the rule. Negation can be allowed in the body, in the head, in both or in none.
noiseThreshold	Maximum proportion of counter-instances of rules. If set to 0, only satisfied rules will be given.
numberLiterals	Maximum number of literals in a rule.
repeatLiterals	Repeated attributes allowed.
rocAnalysis	Return TP-rate and FP-rate for each rule found.
valuesOutput	Give visual feedback during the search. The current best and worst values can be output either to stdout or to a separate window.

The table below describes the capabilites of Tertius.

Capability	Supported
Class	Binary class, Nominal class, Missing class values
Attributes	Unary attributes, Nominal attributes, Missing values, Empty nominal attributes, Binary attributes
Min # of instances	1