Not to include in LUQ. Should just be part of notebooks (and eventually scripts) for now (eventually incorporated into BET).
Also, the command:
print('Average rejection rates for clusters:', rs)
should be changed. I think something like
print('Diagnostic for clusters [sample average of ratios in each cluster]:', rs)
since each r is not technically a rejection rate but rather a relative likelihood update (a re-weighting) to the initial density that can be used to perform rejection sampling.
In the end, we are leaving the compute_r and generate_kdes in here just to run the examples that are in here. These are not referenced by BET though, so it is fine.
Not to include in LUQ. Should just be part of notebooks (and eventually scripts) for now (eventually incorporated into BET).
Also, the command:
print('Average rejection rates for clusters:', rs)
should be changed. I think something like
print('Diagnostic for clusters [sample average of ratios in each cluster]:', rs)
since each r is not technically a rejection rate but rather a relative likelihood update (a re-weighting) to the initial density that can be used to perform rejection sampling.