I have a dataset which consists of two classes. Each record in that dataset is described by a bitvector. I split the dataset to learn and test set both containing items from two classes.
Now I would like to create classification model using Association rules nodes but I'm confused a little bit. Could anyone give me some advice how to do it?
I guess, you can use the normal Association Rule Learner but need to match the sets afterwards with the test data. Please check out the Subset Matcher node which allows finding subsets of cells within a collection. However, you need to translate your bitvector into a collection first - if somehow possible?
Bit vectors can be converted into individual numerical integer columns using the erlwood fingerprint expander node.
I've done the same process, used bitvector before the association rule node but the association rule creates an empty table.
The bitvector variable looks ok, so I'm not sure what I'm doing wrong.
I guess, you need to lower the support (in%) in the node configuration, that is, the minimum number of items that appear frequent within all transactions...
I've tried that too, inicially had it at 0,01 now even 0,0001 still gives an empty table..I've tried increasing too and nothing.
other settings=> underlyning structure=Array; item set type=free;item length=5;confidence=0,75
Any ideas on how to solve this???Thank you
Can you post a snippet of your data, I happy to prepare a small workflow.
Please find attached the example workflow, let me know if this is helpful or not?
Thank you so much!!! I see what i was doing wrong, I must filter all columns that the node doesn´t need for the association rules.
The data I've attached is a small sample and supposed to be ready for association rule.
Thank you so much!!!