#!/usr/bin/env python
# coding: utf-8

# In[2]:


get_ipython().run_line_magic('matplotlib', 'inline')
import numpy as np
import pandas as pd
import pymc3 as pm
import seaborn as sns


# In[3]:


k = 1000
n = 100

fake_data = np.random.random((n, k+1))

y = (fake_data[:, 0] > 0.5).astype(int)
X = fake_data[:, 1:]


# In[4]:


y[:10]


# In[6]:


X[:10]


# In[7]:


import theano.tensor as tt

invlogit = lambda x: 1/(1 + tt.exp(-x))

with pm.Model() as logistic_model:
    
    β = pm.Normal('β', 0, sd=10000, shape=k)
    
    π = invlogit(tt.dot(X, β))
    
    likelihood = pm.Bernoulli('likelihood', π, observed=y)


# In[10]:


with logistic_model:
    tr = pm.sample(10000, step=pm.Metropolis())


# In[11]:


pm.summary(tr[-1000:])