Classification Example with Keras in Python

   In my previous post, I explained how to build Keras deep learning model in R. This post is about how to create a Keras deep learning model in Python. Keras is a high-level neural network API to build deep learning models and can run on top of Tensorflow, CNKT, and Theano. If you don't have Keras and Tensorflow on your work environment, please refer to this link to install them.
   The tutorial cover:

1. Preparing the data
2. Building the model
3. Accuracy check
4. Source code listing

   We'll start by loading the required libraries.

from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from sklearn.metrics import confusion_matrix
from keras.models import Sequential
from keras.layers import Dense

Preparing the data

   We'll use the iris dataset as a target classification data. First, we load the dataset, then separate its feature (x) and label (y) parts, and split it into the train and test parts.

iris = load_iris()
x, y = iris.data, iris.target
xtrain, xtest, ytrain, ytest = train_test_split(x, y, random_state=1)

Building the model

   Next, we build a keras sequential model and fit it with train data. The model is defined by the Keras Sequential() class.

model = Sequential()

Then, we'll add a Dense layer defining the input dimension. We can find out the data dimension by checking its shape. We'll set 4 into the input_dim parameter. Activation will be a relu.

print(xtrain.shape)

(112, 4) 

model.add(Dense(32, input_dim = 4, activation = 'relu'))

We can also add a dropout layer to ignore 20 percent of neurons after the input layer.

model.add(Dropout(0.2))

We'll add one more Dense layer.

model.add(Dense(8, activation = 'relu'))

The final Dense layer identifies the output class number. In iris data, three classes are available, thus we set 3 and activation will be "softmax" type.

model.add(Dense(3, activation = 'softmax'))

We can compile the model with the below parameters.

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

print(model.summary())

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense_22 (Dense)             (None, 32)                160       
_________________________________________________________________
dropout_3 (Dropout)          (None, 32)                0         
_________________________________________________________________
dense_23 (Dense)             (None, 8)                 264       
_________________________________________________________________
dense_24 (Dense)             (None, 3)                 27        
=================================================================
Total params: 451
Trainable params: 451
Non-trainable params: 0
_________________________________________________________________ 

Finally, we'll fit the model with train data.

model.fit(xtrain, ytrain, epochs = 40, batch_size = 16)

After the training of the model, we can evaluate the model accuracy.

acc = model.evaluate(xtrain,ytrain)
print("Loss:", acc[1], " Accuracy:", acc[1:2])

Loss: 0.9196428571428571  Accuracy: [0.9196428571428571]

Accuracy check

The model is ready, now we can predict the test data.

pred = model.predict(xtest)

To transfer pred data into the class id list, we'll use the below function.

pred_y = pred.argmax(axis=-1)

Finally, we'll check the confusion matrix table.

cm = confusion_matrix(ytest, pred_y)
print(cm)

[[13  0  0]
 [ 0 12  4]
 [ 0  0  9]]

   In this tutorial, we have briefly learned how to build Keras model to classify data in Python. The full source code is listed below. Thank you for reading!


Source code listing 

from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from sklearn.metrics import confusion_matrix
from keras.models import Sequential
from keras.layers import Dense, Dropout

iris = load_iris()
x, y = iris.data, iris.target
xtrain, xtest, ytrain, ytest = train_test_split(x, y, random_state=1)

print(xtrain.shape)

model = Sequential()
model.add(Dense(32, input_dim = 4, activation = 'relu'))
model.add(Dropout(0.2))
model.add(Dense(8, activation = 'relu'))
model.add(Dense(3, activation = 'softmax'))

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

print(model.summary())
model.fit(xtrain, ytrain, epochs = 40, batch_size = 16)

acc = model.evaluate(xtrain,ytrain)
print("Loss:", acc[1], " Accuracy:", acc[1:2])

pred = model.predict(xtest)
pred_y = pred.argmax(axis=-1)

cm = confusion_matrix(ytest, pred_y)
print(cm)