Wednesday, June 17, 2015

Wednesday: Today was not the best day ever.

Although I did realize that the reason the one algorithm wasn't giving me results I liked was that it was weighting things by the absolute value instead of the square.  That insufficiently downweighted outliers, producing better but not sufficient set of fits.  Now I can implement that, and do a large scale test.  It's probably actually good enough to skip my other outlier rejection as well.


No comments:

Post a Comment