News for hackers. Stuff that matters.
Re:Linear regression by apathy on Tuesday August 09, 2016 @11:31AM (#12255272)
Eh, orthogonalizing variables is a necessary precondition for this. When was the last time variables were totally uncorrelated and you didn't know it beforehand? Unfortunately such cases are the exception rather than the rule, and you lose interpretability by projecting into a low-rank orthogonal subspace. Not that interpretability is necessarily the be-all and end-all, but the "terms and conditions" are pretty important here!
Re:Linear regression by thanatropism on Tuesday August 09, 2016 @11:47AM (#12255375)
> When was the last time variables were totally uncorrelated and you didn't know it beforehand

Whenever we have an actual theory of the underlying phenomena.

It's common in econometrics to impose orthogonality conditions on the theoretical equations so structural parameters can be recovered; or so we can filter partial correlations. For example: a regression of prices and quantities sold will mix the effect of shifts in demand and in supply. Solution: find something that correlated with demand quantities but not price directly (for example, changes in consumers' wealth).

A realistic, much cooler example:

http://homepage.univie.ac.at/robert.kunst/pan2011_pres_rabas...

Re:Linear regression by smallnamespace on Tuesday August 09, 2016 @01:30PM (#12256239)
> no belief in statistics required

Yes, but now you have a statistical assumption: that your structural model holds.

To make use of the Fritsch-Waugh-Lovell theorem, you must test that your orthogonality assumption holds using statistics :)