Large-scale least-squares regression (was: automate writing formulas)

["Alan Hochberg" <alan.hochberg@prosanos.com>]

Your point is well taken.  When Acton's quote was written, your 300
variables might come from 300 sensors on an aircraft or a chemical reactor,
and there was a chance that physical and engineering insight might tell you
which five or six variables to start with, and which ones might add
information not already included in the model.  Nowadays, the flood of
variables is likely to come from a microarray or an image sensor.  Insight
plays less of a role, and we're left with the strategy of asking the data
which variables to use, which, as you pointed out, is an iffy proposition.

Thanks for the note and the reference.

Alan