Rob Collie uses DAX to find disproportionate sentiment on a per-group basis:
My wife loves to travel AND she loves data, so it’s no surprise that she showed me the infographic in question.
“Oh come on, California residents don’t visit the Philippines more often than any other country.” That was my first reaction. (Mexico is the overwhelming #1 destination, basically for every state… because Cancun. And because there are so many Mexican-Americans).
“Ah… but California residents DO visit the Philippines disproportionately more often than they ‘should,’ according to national averages!” That was my dawning second realization (and confirmed by the fine print of the Orbitz article, even though the article’s title suggests otherwise.)
So, how do we do that in DAX? Pretty simply, actually.
I’m mentally working out whether this could be useful in anomaly detection.