I want to find frequent item sets with the itemset finder or/and the association rule learner.
My problem is that they are generating “wrong” results. I think the format of my input data is wrong.
I am using the same structure as in the example workflows about association rules. My items are separated by “,” without any spaces and they are within a collection.
I noticed the error at support:
One of my products has 100 percentage support. When i use Row Filter to select these rows, I get all rows (so support is 100%). Itemset Finder finds the itemset, yes, but only with a support of 6%.
Could you post a sample workflow or dataset so that I might assist? If you are dealing with proprietary data, perhaps some dummy rows would suffice? That would help me have a better understanding of the problem you’re running into.
I have still the same problem with KNIME 4.0.2. It seems that items which have 100% support never appear in itemsets with 2 or more elements. I am attaching a workflow which shows this problem. The association rule learner node should produce as a result all the combinations of item0, item1 and item2, but it only shows item0 alone.
I see what you’re talking about here. It doesn’t line up with what I’d expect either.
We’re going to look into it.
In the mean time the spark frequent item sets node is outputting correctly. Although it does require a slightly different input configuration it might be a temporary work around.
Hello gamato,
please use the Association Rule Learner (Borgelt) which outputs the correct results (see attached workflow). We will look into the issue with the other Association Rule Learner node.
Bye
Tobias
