Thank you very much for the reply.
To understand further about the custom distribution class configuration:
- What are the necessary components needed to design the class custom distribution?
- Is it necessary to include the
def random
in the custom distribution? - What is the reason for getting different fused distributions when implementing the bayesian fusion part? Could it be because of
def random
code section in the custom distribution? For clarification about the fusion part, please find the following question: How to create Bayesian data fusion in python with pymc3?.