[twic]

I'd be interested to see plots of median or upper-quartile inflation across the CPI categories in there too. In periods of high inflation, is the high variation really random variation about an increased centre, or does it reflect the departure of an extreme from an unchanged centre?

In the article and my suggestion, there's also an unexamined assumption here that the CPI categories are peers, on which it's valid to do statistics. It's not clear to me that "medical care commodities" and "shelter" are really things it makes sense to take an arithmetic mean of.