Weightened linear model in GLM.jl

[achjaj]

Hi,

I want to fit a plane to measured points, i.e., the function Z = aX + bY + c to data, where the Z coordinate (dependent variable) has a measurement error \sigma. The documentation for GLM.jl mentions that lm function can take a vector of weights, but I’m not sure what these weights are supposed to be. Can I simply use 1/\sigma^2 as the weights, similar to LsqFit.jl?

[nilshg]

IIRC weights in GLM are frequency weights, see an old discussion here:

Usage of different types of weights

The weights.jl file describes three types of weights: frequency weights, probability weights, and analytic weights. This is an amazing feature to Julia, as only commercial software like STATA and SAS understand the differences between these 3 weights. R and Python only understand one type of weight, which I think is something like an importance weight. Being able to use these 3 types of weights properly is crucial to the field that I work in (causal machine learning). Most software implementat…

[achjaj]

Thanks for your reply. Can you please elaborate more? Frequency of what?

[nilshg]

There’s a set of lecture notes on the linked thread that explains frequency, probability and analytical weights. In short, frequency here refers to the frequency of an observation.