How to filter trivial/redundant association rules


i am mining association rules from sales data across many taxonomies.

Two examples for taxonomies are Clothes -> T-Shirt -> Blue T-Shirt or Shoes -> Summer Shoes --> Summershoe with Size X.

The lowest taxonomy (Blue shirt or shoes with specific size) is listed within the transactions. I extended all transactions with the whole hierarchy. E.g. someone bought the blue shirt, his transaction also contains “T-Shirt” and “Clothes”. Consequently I get good rules over all hierarchies.

My problem now is that there are also rules like “T-Shirt -> Clothes”. This is in some way trivial because its between the same hierarchy.

Do you have an idea how to filter those rules, who include a taxonomy? E.g. with a dictionary that represents the “is-a” taxonomy?

maybe the rulebased Rowfilter Node helps…

there you exclude the Rules of your choice…
use the RuleID for it and put a querry like:
$RuleId$ LIKE “Row10” OR $RuleId$ LIKE “Row11” => TRUE
then activete the “Exclude True matches” below

1 Like