support for incremental computation of association rules

[rjoberon]

The computation of association rules that fulfill certain minimal support and confidence constraints could be done with algorithms that are particularly suited for that, e.g., TITANIC, FP-Growth and the like. This would be more efficient than first computing all rules and then restricting them.

Currently this is out of scope of ConExp-NG and other tools exist for that (e.g., Weka). However, in the future this might be an option to add.

related to issue #135