I can't. Too much proprietary info.
My data would look like this:
Production Data :
PROD_ DATE, Model number, SERIAL
3/1/17 2:42pm, A, 1001
3/1/17 2:43pm, A, 1002
3/1/17 2:44pm, B, 1001
3/1/17 2:45pm B, 1002
3/2/17 2: 46pm, C, 1001
DEFECT_DATE, Model Number, Serial
3/5/17, A, 1001
3/6/17, A, 1002
I want to count the unique serials for each model per desired production time period (day, week, month, etc).
Similar for the defects.
Model, Month, Defect Count, Production Count, (Defect Count/Production count*100)
Also show in a graph:
Yr_Month 2017_1, 2017_2, 2017_3, ...
Model A Defect count/Production count*100 for each month
Model B Defect count/Production count*100 for each month
Note: I don't need to join anything by serial number. I just need to count the unique serials in a given time period.
I have studied your article and other set analysis videos and think I have made good progress.
For each year I have created the following expression to generate my quality ratio:
ie: Ratio= (Count total defects for 2013)/(Count total serial numbers for 2013)
Are there any problems with dividing one set expression for "defect count" by the other set expression "serial number count"? They come from different tables.
Everything seems to filter properly using normal selections.
Note: I ran into much trouble trying to filter the year using derived calendar fields (still a known bug?) so I created a new field using a date function in the import script. Now my filters work.