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

# Gaussian Naive Bayes
# ====================

# Let's set some setting for this Jupyter Notebook.

# In[2]:


get_ipython().run_line_magic('matplotlib', 'inline')
from warnings import filterwarnings
filterwarnings("ignore")
import os
os.environ['MKL_THREADING_LAYER'] = 'GNU'
os.environ['THEANO_FLAGS'] = 'device=cpu'    
    
import numpy as np
import pandas as pd
import pymc3 as pm
import seaborn as sns
import matplotlib.pyplot as plt
np.random.seed(12345)
rc = {'xtick.labelsize': 20, 'ytick.labelsize': 20, 'axes.labelsize': 20, 'font.size': 20, 
      'legend.fontsize': 12.0, 'axes.titlesize': 10, "figure.figsize": [12, 6]}
sns.set(rc = rc)
from IPython.core.interactiveshell import InteractiveShell
InteractiveShell.ast_node_interactivity = "all"


# Now, let's import the `GaussianNB` (GaussianNaiveBayes) model from the `pymc-learn` package.

# In[3]:


import pmlearn
from pmlearn.naive_bayes import GaussianNB
print('Running on pymc-learn v{}'.format(pmlearn.__version__))


# ## Step 1: Prepare the data
# Use the popular iris dataset.

# In[4]:


# Load the data and split in train and test set
from sklearn.datasets import load_iris
X = load_iris().data
y = load_iris().target
X.shape


# In[5]:


from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3)


# ## Step 2: Instantiate a model

# In[6]:


model = GaussianNB()


# ## Step 3: Perform Inference

# In[7]:


model.fit(X_train, y_train, minibatch_size=20, inference_args={'n': 60000})


# ## Step 4: Diagnose convergence

# In[8]:


model.plot_elbo()


# In[9]:


pm.traceplot(model.trace);


# In[10]:


pm.forestplot(model.trace);


# ## Step 5: Critize the model

# In[11]:


pm.summary(model.trace)


# In[12]:


pm.plot_posterior(model.trace);


# ## Step 6: Use the model for prediction

# In[13]:


y_probs = model.predict_proba(X_test)


# In[14]:


y_predicted = model.predict(X_test)


# In[15]:


model.score(X_test, y_test)


# In[16]:


model.save('pickle_jar/gaussian_nb')


# #### Use already trained model for prediction

# In[17]:


model_new = GaussianNB()


# In[18]:


model_new.load('pickle_jar/gaussian_nb')


# In[19]:


model_new.score(X_test, y_test)


# ## MCMC

# In[20]:


model2 = GaussianNB()
model2.fit(X_train, y_train, inference_type='nuts')


# ### Diagnose convergence

# In[21]:


pm.traceplot(model2.trace);


# In[22]:


pm.gelman_rubin(model2.trace)


# In[23]:


pm.energyplot(model2.trace);


# ### Criticize the model

# In[24]:


pm.summary(model2.trace)


# In[25]:


pm.plot_posterior(model2.trace);


# ### Use the model for prediction

# In[26]:


y_predict2 = model2.predict(X_test)


# In[27]:


model2.score(X_test, y_test)


# In[28]:


model2.save('pickle_jar/gaussian_nb2')
model2_new = GaussianNB()
model2_new.load('pickle_jar/gaussian_nb2')
model2_new.score(X_test, y_test)