associations - Weka Apriori No Large Itemset and Rules Found -

i trying apriori association mining weka (i use 3.7) using given database table

so, exported 2 columns (orderlinenumber , productcode) , load weka, far go, haven't got success attempt, ended "no large itemsets , rules found!"

again, tried convert csv arff file first using arff converter , still same message;

i tried using database loader in weka, data loaded fine still give same result;

the filter i've applied in preprocessing numerictonominal filter;

what have wrongly done here, suspiciously think arff format though, thank you

update after further trial, found out exported wrong column , lack 1 filter process, "denormalized", installed plugin via packet manager , denormalized data after converting nominal first;

i compared results "supermarket" sample's result; difference output came 'f' instead of 't' (like shown below) , confidence value seems 100%;

first of all, orderline wrong column.

obviously, position on printed bill not important.

secondly, file format not appropriate.

you want one line every order, 1 column every possible item in @data section. save memory, may helpful use sparse formats (do not forget set flags appropriately)

other tools elki can process input formats this, may easier use (it lot faster weka):

apple banana milk diapers beer 

but last checked, elki "only" find frequent itemsets (the harder part) not compute association rules. used tiny python script produce actual association rules desired.