##Apriori Tensorflow
Apriori implementation using tensorflow backend
- Test with a large sample set to check performance against e.g. apyori
- This shows the memory usage is a big issue - even 100'000 inputs with 30 possible values surpassed 100Gb (95Gb compressed) usage. Need to remove rows which are no longer possible. The lack of sparse element wise multiplication is now a blocker to this project.
- Neaten up output - return AprioriFrequentSets
- Put apriori.py functions into classes, store result as member for further processing
- Add feature to return list of all 'members' (input row ids) in each frequent group
- Add feature to calculate distance metric between groups