Background: I’m trying to fit a 2D symmetric Gaussian to a an image of a point source. So I have an image of a light source, it’s Gaussian in shape, and I want to find it’s exact center and get a measure of how good the fit is. I plan to
log the intensities and fit a symmetric paraboloid to the data. To compensate for the over represented tails of the Gaussian, I want to weight the data with itself (so low intensities will have low weights and vice versa). Here’s what it looks like (artificially enlarged):
Question: In order to fit a linear model to weighted data I assume I need to use
MultivariateStats.llsq, but I can’t really see how. What is the way to fit multiple linear regression to weighted data in Julia?
I found tons of functions in
StatsBase that act on the subtypes of
RegressionModel, but not one function that produces any instances of its subtypes…