Specifying loc and scale on beta prior requires bounds, gives error

Questions version agnostic
3

Sure, it’s like this:

with pm.Model() as model1:
    bounded_laplace = pm.Bound(pm.Laplace, lower=0.0)
    alpha = bounded_laplace('alpha', mu = 51704326, b = 767593)

    beta_param = SSBeta(name = 'beta_testing',
                    alpha = 1.578, 
                    beta = 0.855, 
                    loc = 0.0004,
                    scale = 7.338e-5)
    pred = pm.Gamma("test_var", alpha = alpha, beta = beta_param, 
    observed=np.array([1.72e6, 1.58e6, 1.64e6, 1.59e6]))

    trace = pm.sample(400)
1 Like