"""
Use the naive bayes classifier to classify the digits data set.

- This is an example of a supervised ML algorithm
    - it has labels on the training data                                                    
    - you tell the model: this is class X during training
"""
import matplotlib.pyplot as plt
import numpy as np

from sklearn import datasets
from sklearn.model_selection import train_test_split
from sklearn.naive_bayes import GaussianNB

digits = datasets.load_digits()
print(digits.data.shape)

# split into training and test data
x_train, x_test, y_train, y_test = train_test_split(
    digits.data, digits.target, test_size=0.2, random_state=0
)

# use a gaussian NB classifier
classifier = GaussianNB()
# train on the split data
classifier.fit(x_train, y_train)
# test the model and print it's accurecy
score = classifier.score(x_test, y_test)
print(score)

# visualizing the learned means as 8x8 images
fig, axes = plt.subplots(2, 5, figsize=(12, 5))
for i, ax in enumerate(axes.flat):
    ax.imshow(classifier.theta_[i].reshape(8, 8), cmap='gray_r')
    ax.set_title(f'Class {i}')
    ax.axis('off')
fig.suptitle('NB: Mean pixel intensity per class')
fig.savefig('naivebayes_digits_means.png', dpi=150, bbox_inches='tight')

# The variance plot shows where pixels vary most within a class: 
# - high variance (bright) means that pixel isn't reliable for classification
# - low variance (dark) means it's consistent.
fig, axes = plt.subplots(2, 5, figsize=(12, 5))
for i, ax in enumerate(axes.flat):
    ax.imshow(classifier.var_[i].reshape(8, 8), cmap='hot')
    ax.set_title(f'Class {i}')
    ax.axis('off')
fig.suptitle('NB: Pixel variance per class')
fig.savefig('naivebayes_digits_variance.png', dpi=150, bbox_inches='tight')

# plot the variance difference between two commonly confused digits like 3 and 8 to see
# on which pixels nb relies to tell them apart
plt.figure()
diff = abs(classifier.var_[3] - classifier.var_[8])
plt.imshow(diff.reshape(8, 8), cmap='hot')
plt.title('Variance difference: 3 vs 8')
plt.colorbar()
plt.savefig('naivebayes_3v8_variance.png', dpi=150, bbox_inches='tight')