Are there any common approaches that split the bins into non-equal sizes? If so, what are some common other ways to split them?
How do we determine the size of the bins?
How do you split a potentially non-uniform space into uniform bins? Also, why are there gaps in uniform sampling to begin with? Why don't we just pick points from a normal gaussian with lower variance?
Are there types of target distributions where stratified sampling would not work/be as effective in reducing variance?
How many bins is a good amount in practice?
Is there a certain strategy we can apply to all cases to determine size and number of bins? Or is it dependent on the situation?
Although it has already been asked above, I also want to know how bin width should be determined for stratification!
How do we know what bin width is good; and given a bin width and sampling, how do we know if the sample is good enough?