I am currently having a dataset with several test cases which describes passes of tests (1), failures of tests (0) and in case that a test case has never been done (None).
My aim is to determine the correlations/associations so that I can provide some insights on which parts of the data, the more frequent failures occur as well as for passes.
I came across the apriori/fpgrowth algorithms, although they are suitable to determine frequencies as if a pass/failure was a single item in the dataset. Hence, it only computes the corresponding metrics based on the existence of pass test cases, neglecting the rest as 'non-existent' (Failures and None).
My question is if it is possible to use these algorithms with all the cases I am having on the dataset (eg. Pass, Failure, None). So that I can determine if there's influence/correlation between pass test cases and fail test cases.
I only came across with the idea of implementing the algorithm two separate times (one for pass test cases, and the other for fail) and obtain results for two separate cases. But I wasn't sure if this will be sensible enough for my final results.
Some help on this complicated problem would be very much appreciated. :)
