From my perspective the attached file works; I've made a couple of changes to Detailed table 2 and added a table box with the output of "production table"
You still have the issue with Sept-2017 but it is related to the data model: neither of the two tables in your data set have primary keys. I think this creates confusion when you look at the data. You have the same transaction multiple times in the table Production_Defects and multiple times in the table "test". This wouldn't necessarily be a problem but in your case you are using values from these tables which might very.
Let me give you an example: if you look at transaction "SSDRC-391" this changes severity (as per table test): it starts with a severity of 2 and then, on the 11th of October goes up to 3. This creates an issue in your visualisation in "Test 2" chart: there is an "orphan" severity 2 incident that gets flagged when looked into the context of the severity 2 column but disappear in the Detailed Table2 (the one that I've created) because that entry in its entirety (i.e. without looking at severity) is not causing one of the discrepancies you want to analyse.
The other issue that you have is that some transactions have a blank date (or are missing) from the test table. In your Test 1 chart you use a dimension (PDReportPeriodB) coming from Production_Defects table. In the Test 2 chart you use the same dimension but you apply a set analysis based on a field in the "test" table; where that field is null the set analysis will return nothing (i.e. that transaction cannot be evaluated). The issue arises when you try to extract the information of the missing transaction and try to apply a filter on a field which is null; it doesn't work as that field for that line in the database does not exist.
I think you should do (or explain so that we can help) two things:
1. explain how your data model is structured, what info it contains and how they are related to one another
2. explain what you need to visualise; what is the business requirement
Lorenzo
