Thank you so much! I finally understand how to distinguish each distributions and their properties and functions! This is very helpful, thanks for your work!!
I'm a bit confused when to do a GLM with a log link vs just a log transform of Y. 11:29 suggests that if data have > variance with > mean, use log-transform of Y. However, the chart at 19:53 says that log link is appropriate when variance ~ mean, which is just another way of saying that variance gets bigger than mean, which contradicts the first point. What am I missing?
