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

# In[1]:


from pycaret.classification import *
from pycaret.datasets import get_data
import pandas as pd


# ## Getting the dataset

# In[2]:


#get dataset
dataset = get_data('bank')


# In[3]:


#check the shape of data
dataset.shape


# In[5]:


dataset.isna().sum()


# In[6]:


## sample returns a random sample from an axis of the object. That would be 38,429 samples, not 45211
data = dataset.sample(frac=0.85, random_state=456)


# In[7]:


data


# In[8]:


# remove from the original dataset this random data
data_unseen = dataset.drop(data.index)
data_unseen


# In[9]:


# Reseting the index of both datasets
data.reset_index(inplace=True, drop=True)
data_unseen.reset_index(inplace=True, drop=True)
print('Data for Modeling: ' + str(data.shape))
print('Unseen Data For Predictions: ' + str(data_unseen.shape))


# In[10]:


model_setup = setup(data=data, target='deposit', session_id=321)


# ## Compare Model

# In[11]:


#Compare Model

best_model = compare_models()


# In[12]:


print(best_model)


# ## Create the Model

# In[13]:


models()


# In[14]:


gbc = create_model('gbc')


# In[15]:


#trained model object is stored in the variable 'gbc'. 
print(gbc)


# ## Tunning the Model

# In[17]:


# Accuracy in previous model is 0.9056 and in the tuned model accuracy is 0.9065

tuned_gbc = tune_model(gbc)


# In[18]:


print(tuned_gbc)


# ## Plotting the Model

# In[19]:


## AUC Plot

plot_model(tuned_gbc, plot = 'auc')


# In[20]:


## Consufion matrix

plot_model(tuned_gbc, plot = 'confusion_matrix')


# ## Evaluation the model
# 

# In[21]:


## model performance is to use the evaluate_model()
evaluate_model(tuned_gbc)


# ## Finalizing the Model

# In[22]:


final_gbc = finalize_model(tuned_gbc)
final_gbc


# In[23]:


#Final gbc parameters for deployment
print(final_gbc)


# ## Predicting with the model

# In[24]:


predict_model(final_gbc)


# In[25]:


unseen_predictions = predict_model(final_gbc, data=data_unseen)
unseen_predictions.head()


# ## Save Model 

# In[26]:


save_model(final_gbc, './Pycaret/Final_gbc')


# ## Load Model

# In[27]:


saved_final_gbc = load_model('./Pycaret/Final_gbc')


# In[28]:


new_prediction = predict_model(saved_final_gbc, data=data_unseen)


# In[29]:


new_prediction.head()


# In[ ]: