I have a fact table that has several fields that are Y/N in the source system.
Originally, I created the fact table converting the data from Y to 1 and N to zero, with the measure aggregate function property set to "sum".
However after reading some of the posts by Mosha, I think I should be converting these N's to NULLs instead of Zero's to take advantage of NON EMPTY.
Is my logic correct?
It really depends on what you want to do with this data. It's hard to give advice without more details. Typically, in the questionarie type of models, the Y/N columns are converted to attributes, not to measures, and the measure is count. This way it is easy to get how many people answered Yes and how many people answered No. By doing this as measure - you will have easy way to find out how many said Yes, but you will lose info about how many said No.|||In this case the yes / no indicate if a particular production step was performed. So the values will get summed up for a total.
the no's mean nothing to me.
that total may then also be used to determine a percent of all products that had that production step performed.
|||Then it is possible to model it the way you proposed (although I still think that making a Y/N attribute is a good idea).