Just a general question on whether to create new columns in Fact Tables or create measures.
Am I correct in thinking that by creating columns it creates a single overhead at refresh time but would measures not potentially increase the “in line” processing overhead?
Could it be that under certain circumstances creating a column would be more advisable? Particularly where there may be a structure of “nested” measures?
Calculated Columns will increase the file size and performance also stored in RAM. Measures do not consume RAM but CPU. They are benefits to each, depending on the use. Have a read here for more of an explanation.
I always look at calculated columns as a last resort. If something needs to be a column (e.g., you want to be able to slice on it), I try to do that in Power Query first, for the reasons @marcster_uk cites above. Personally, I also find it a bit disorienting going into PQ in my report development and not seeing all the relevant columns in a table (since calculated columns will not show up there).