"""
MNIST classification with a fully-connected (dense) neural network.

Each 28x28 image is flattened into a linear list of 784 inputs
"""
######### COPY PASTE from import ... line to (x_train, ... line
##########################################################################
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers

EPOCHS = 5
BATCH_SIZE = 128

tf.random.set_seed(0)

# Load data and scale pixel values to [0, 1]
(x_train, y_train), (x_test, y_test) = keras.datasets.mnist.load_data()
##########################################################################

##########
###### TYPE IN THE NEXT 4 individual lines and then the 3 lines that show one example
###### and then change the 0 to 10 in the ones you typed to see different example
##########

# Get a feel for the data you have loaded in type the following single lines in and
# observe the output:

####
x_train.shape 
#shows 60000 rows of training examples, each with 784 values (the gray-scale pixel values)

####
y_train.shape
#shows 60000 values which are the labels of each of the 60000 training examples

####
x_train[0]
#shows the 784 pixel values note: that many are 0 and others are close to 255,
# you will need to scroll to see non-zeros

####
print(y_train[0])
#shows the label of the 0th training image which is 5

# this next chunk will show the image of the 0th training example which is a handwritten 5
####
import matplotlib.pyplot as plt
plt.imshow(x_train[0], cmap="binary")
plt.axis("off")
plt.show()
# show the image of the 0th training example which is a handwritten 5

# go back to x_train[0] cell and below and instead of looking at the 0th 
#training example and its label enter say 10
#so x_train[10], y_train[10] and in imshow(x_train[10] etc.

##########################################################################

######### COPY PASTE next 2 lines which convert the input values from 0-255 to 0-1
##########################################################################
x_train = x_train.astype("float32") / 255.0
x_test = x_test.astype("float32") / 255.0
##########################################################################

##########
###### TYPE IN THE NEXT 2 individual lines
##########

# The two lines above converted the values from 0-255 to decimal values from 0-1
# notice:
x_train[10]
# now has decimal values rather than whole numbers from 0 to 255

# see the shape of one training example:
x_train[10].shape
# it is two dimensional 28 rows x 28 columns

######### COPY PASTE next 4 lines which flatten each 28x28 image into 784 values
##########################################################################
x_train = x_train.reshape(-1, 28 * 28)
x_test = x_test.reshape(-1, 28 * 28)
print("Training input shape:", x_train.shape)   # (60000, 784)
print("Test input shape:    ", x_test.shape)    # (10000, 784)
##########################################################################

# the above lines flatten the 2 dimensional images into 1 dimension of 784 values each
# so now
x_train[10].shape
# is 1 dimension of 784 values

######### COPY PASTE next 11 lines which sets up the artifical neural network
##########################################################################
model = keras.Sequential([
    keras.Input(shape=(784,)),
    layers.Dense(8, activation="relu"),
    layers.Dense(10, activation="softmax"),
], name="dense_mnist")

model.summary()

model.compile(optimizer="adam",
              loss="sparse_categorical_crossentropy",
              metrics=["accuracy"])
##########################################################################

######### COPY PASTE from model.fit to second print which does the training and then testing
##########################################################################
model.fit(x_train, y_train,
          epochs=EPOCHS,
          batch_size=BATCH_SIZE,
          validation_split=0.1,
          verbose=2)

test_loss, test_acc = model.evaluate(x_test, y_test, verbose=0)
print(f"\nDense NN parameters: {model.count_params():,}")
print(f"Dense NN test accuracy: {test_acc:.4f}")
##########################################################################