Experimental Designs for Heteroskedastic Variance

Accurate data analysis pipelines require decomposing observed data into signal and noise components. To facilitate this, the noise is commonly assumed to be additive and homogeneous.