Open bethac07 opened 2 years ago
was it ever considered to use CV rather than pure standard deviation in this feature?
I don't believe so. @ruifanp developed the approach, so he might have some more insight.
In general, I think that we're definitely open to adding functionality here!
Purely a question - was it ever considered to use CV rather than pure standard deviation in this feature? This is an obviously artificial example, but to me I would feel 'b' is more noisy in that across my feature the fold change is really large, but since 'a' has a higher absolute value it gets removed first