Chapter 4 Hierarchical linear regression error

[chdamianos]

I get an error in chapter 4 when I try to run the code for the hierarchical linear regression model

with pm.Model() as hierarchical_model:
    # hyper-priors
    alpha_tmp_mu = pm.Normal('alpha_tmp_mu', mu=0, sd=10)
    alpha_tmp_sd = pm.HalfNormal('alpha_tmp_sd', 10)
    beta_mu = pm.Normal('beta_mu', mu=0, sd=10)
    beta_sd = pm.HalfNormal('beta_sd', sd=10)

    
    # a prioris
    alpha_tmp = pm.Normal('alpha_tmp', mu=alpha_tmp_mu, sd=alpha_tmp_sd, shape=M)
    beta = pm.Normal('beta', mu=beta_mu, sd=beta_sd, shape=M)
    epsilon = pm.HalfCauchy('epsilon', 5)
    nu = pm.Exponential('nu', 1/30)

    y_pred = pm.StudentT('y_pred', mu=alpha_tmp[idx] + beta[idx] * x_centered, sd=epsilon, nu=nu, observed=y_m)

    alpha = pm.Deterministic('alpha', alpha_tmp - beta * x_m.mean()) 
    alpha_mu = pm.Deterministic('alpha_mu', alpha_tmp_mu - beta_mu * x_m.mean())
    alpha_sd = pm.Deterministic('alpha_sd', alpha_tmp_sd - beta_mu * x_m.mean())
    
    mu, sds, elbo = pm.variational.advi(n=100000, verbose=False)
    
    cov_scal = np.power(hierarchical_model.dict_to_array(sds), 2)
    step = pm.NUTS(scaling=cov_scal, is_cov=True)
    trace_hm = pm.sample(1000, step=step, start=mu,njobs=1,chains=1)

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-55-354ef54155e7> in <module>()
     20 
     21     # line wrong, cannot find solution
---> 22     mu, sds, elbo = pm.variational.advi(n=100000, verbose=False)
     23 
     24     cov_scal = np.power(hierarchical_model.dict_to_array(sds), 2)

AttributeError: module 'pymc3.variational' has no attribute 'advi'

These are the details of my installation

Python 3.5.5 IPython 6.4.0 PyMC3 3.4.1 NumPy 1.14.3 SciPy 1.1.0 Matplotlib 2.2.2 Seaborn 0.8.1

[chdamianos]

Sorry this was meant to be posted in another book's issues page. Anyway, in case somebody ends up here looking for the answer I solved my problem by changing the code to:

with pm.Model() as hierarchical_model:
    # hyper-priors
    alpha_tmp_mu = pm.Normal('alpha_tmp_mu', mu=0, sd=10)
    alpha_tmp_sd = pm.HalfNormal('alpha_tmp_sd', 10)
    beta_mu = pm.Normal('beta_mu', mu=0, sd=10)
    beta_sd = pm.HalfNormal('beta_sd', sd=10)

    
    # a prioris
    alpha_tmp = pm.Normal('alpha_tmp', mu=alpha_tmp_mu, sd=alpha_tmp_sd, shape=M)
    beta = pm.Normal('beta', mu=beta_mu, sd=beta_sd, shape=M)
    epsilon = pm.HalfCauchy('epsilon', 5)
    nu = pm.Exponential('nu', 1/30)

    y_pred = pm.StudentT('y_pred', mu=alpha_tmp[idx] + beta[idx] * x_centered, sd=epsilon, nu=nu, observed=y_m)

    alpha = pm.Deterministic('alpha', alpha_tmp - beta * x_m.mean()) 
    alpha_mu = pm.Deterministic('alpha_mu', alpha_tmp_mu - beta_mu * x_m.mean())
    alpha_sd = pm.Deterministic('alpha_sd', alpha_tmp_sd - beta_mu * x_m.mean())
    
    trace_hm = pm.sample(1000, advi_map=True, chains=1, njobs=1)

[Ernnnn]

Sorry this was meant to be posted in another book's issues page. Anyway, in case somebody ends up here looking for the answer I solved my problem by changing the code to:

with pm.Model() as hierarchical_model:
    # hyper-priors
    alpha_tmp_mu = pm.Normal('alpha_tmp_mu', mu=0, sd=10)
    alpha_tmp_sd = pm.HalfNormal('alpha_tmp_sd', 10)
    beta_mu = pm.Normal('beta_mu', mu=0, sd=10)
    beta_sd = pm.HalfNormal('beta_sd', sd=10)

    
    # a prioris
    alpha_tmp = pm.Normal('alpha_tmp', mu=alpha_tmp_mu, sd=alpha_tmp_sd, shape=M)
    beta = pm.Normal('beta', mu=beta_mu, sd=beta_sd, shape=M)
    epsilon = pm.HalfCauchy('epsilon', 5)
    nu = pm.Exponential('nu', 1/30)

    y_pred = pm.StudentT('y_pred', mu=alpha_tmp[idx] + beta[idx] * x_centered, sd=epsilon, nu=nu, observed=y_m)

    alpha = pm.Deterministic('alpha', alpha_tmp - beta * x_m.mean()) 
    alpha_mu = pm.Deterministic('alpha_mu', alpha_tmp_mu - beta_mu * x_m.mean())
    alpha_sd = pm.Deterministic('alpha_sd', alpha_tmp_sd - beta_mu * x_m.mean())
    
    trace_hm = pm.sample(1000, advi_map=True, chains=1, njobs=1)

for now ,the code can not be run. and I run this code which work for me.trace_hm = pm.sample(1000, init='advi_map', chains=1, njobs=1) Is that right?