And then you have the impact of oil, whose volatility is derived from difficult-to-predict geopolitics and abundance, making it useless as a short-term inflation indicator and even more useless as a very long term indicator, and yet we heavily rely on it regardless. Might as well include Bitcoin while we're at it. Not to mention the absence of real estate / college tuition / credit card debt in these measurements. Here's an interesting look at some cost comparisons, showing just how difficult it is to capture inflation using a CPI or even PCE approach.
http://www.mybudget360.com/cost-of-living-compare-1975-2015-...
While the author is concerned with meeting his distorted inflation metric on paper before raising rates, there's another reason to keep rates low: the cost of debt is about to rise for the world's largest debtor, and the payment is going to further add to its deficit. If the Fed is only concerned with the short-term problems of the US economy, like it usually is, then it should probably never raise rates.