Added Comments to Deep Neural Networks with Keras

Added Comments to Deep Neural Networks with Keras

Author: Manthan Thakker

DeepNeuralNetPython/NeuralNetworkWithKeras.py

@@ -5,9 +5,11 @@
 from keras.optimizers import RMSprop
 import pandas as pd
 
+## 1. Import File
 data = pd.read_csv("/Users/trailbrazer/Desktop/udemy/DataScience-Python3/mammographic_masses.data.txt", na_values=['?'],
                    header=None)
 
+## Preprocessing using Pandas
 data.dropna(inplace=True)
 
 trainingData = data.iloc[:700, 0:4]
@@ -18,26 +20,35 @@
 
 (mnist_train_images, mnist_train_labels), (mnist_test_images, mnist_test_labels) = (trainingData, trainingLabel), (testingData, testingLabel)
 
+#Categorical Labels: For 0 and 1 would be 2. Number of Categories
 train_labels = keras.utils.to_categorical(mnist_train_labels, 2)
 test_labels = keras.utils.to_categorical(mnist_test_labels, 2)
 
+# Model Boiler Plate from Keras, TensorFlow
 model = Sequential()
+#Input Layer Must match the input features
 model.add(Dense(5, activation='relu', input_shape=(4,)))
 model.add(Dense(12, activation='relu'))
 model.add(Dense(2, activation='softmax'))
+#Output Categories = 2 must match the number of categories
+#Activation Function = Relu, Softmax is nothing but determining what ouput => classifier
 
+# Model Sumaary 
 model.summary()
 
 model.compile(loss='categorical_crossentropy',
               optimizer=RMSprop(),
               metrics=['accuracy'])
 
+# Start the training
 history = model.fit(trainingData, train_labels,
                     batch_size=100,
                     epochs=1000,
                     verbose=2,
                     validation_data=(testingData, test_labels))
 
+#Evaluates the prediction
 score = model.evaluate(mnist_test_images, test_labels, verbose=0)
+# Accuracy and loss
 print('Test loss:', score[0])
 print('Test accuracy:', score[1])