We are looking at the behaviour of treatment pathways in order to implement a version running against a spark cluster.

There are a number of results that we are not sure are what was intended.

These are outlined in the image below, where 100, 200 etc are person ids.

The columns are:

1. Input events - event cohorts - assume all these events fall within target
2. Current output - these are current pathways
3. Required output - these are the results we expected
4. Henry has modified runPathways SQL and these are the outputs following his changes. He can comment on this.

I'll add the test data in a follow up comment