Suppose I have a simple model, y=a*x+b I want to model a =pymc3.Normal('a',mu=0,sd=1.,shape=(1)), a normal distribution, and I want to make b as an unknown constant. Then how to assign a pymc3 object to b? Thank you.
Asked
Active
Viewed 109 times
0
-
If you don't know the value of `b` then it should be a probability distribution. Everything you do not know in a Bayesian analysis is represented by a probability distribution. – aloctavodia Jan 08 '18 at 15:06
-
Do you want to get a point-estimate of `b`? In that case, you can compute the posterior and then just report the mean (or another summary statistic) of `b`. – aloctavodia Jan 09 '18 at 14:25