You can control the nature or the number of rules that
are found by the Associations mining function by using rule filters.
Rule filters are a powerful way to limit the amount of rules to be
generated or the content of the rules.
If you do not want to use rule filters, you can use methods for
the most frequently used rule filter constraints to limit the following
rule properties:
- Maximum rule length
- This parameter determines the maximum number of items that occur
in an association rule.
For example, if you specify 3 as the maximum
rule length, the association rules include at most two items in the
rule body and one item in the rule head. The generated rule might
look like this:
[Swimsuit][Beach towel] => [Sunglasses]
- Minimum support
- This parameter determines the minimum threshold for the support
of the rules to be included in the generated rule model.
The support
of a rule indicates the percentage of the input data the rule applies
to.
- Minimum Confidence
- This parameter determines the minimum threshold for the confidence
of the rules to be included in the generated rule model.
The confidence
of a rule indicates how likely it is that an input data record supports
the rule head if it is known that the data record supports the rule
body.