1#  Copyright 2016 The TensorFlow Authors. All Rights Reserved.
2#
3#  Licensed under the Apache License, Version 2.0 (the "License");
4#  you may not use this file except in compliance with the License.
5#  You may obtain a copy of the License at
6#
7#   http://www.apache.org/licenses/LICENSE-2.0
8#
9#  Unless required by applicable law or agreed to in writing, software
10#  distributed under the License is distributed on an "AS IS" BASIS,
11#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12#  See the License for the specific language governing permissions and
13#  limitations under the License.
14"""Example of DNNRegressor for Housing dataset."""
15
16from __future__ import absolute_import
17from __future__ import division
18from __future__ import print_function
19
20import numpy as np
21from sklearn import datasets
22from sklearn import metrics
23from sklearn import model_selection
24from sklearn import preprocessing
25
26import tensorflow as tf
27
28
29def main(unused_argv):
30  # Load dataset
31  boston = datasets.load_boston()
32  x, y = boston.data, boston.target
33
34  # Split dataset into train / test
35  x_train, x_test, y_train, y_test = model_selection.train_test_split(
36      x, y, test_size=0.2, random_state=42)
37
38  # Scale data (training set) to 0 mean and unit standard deviation.
39  scaler = preprocessing.StandardScaler()
40  x_train = scaler.fit_transform(x_train)
41
42  # Build 2 layer fully connected DNN with 10, 10 units respectively.
43  feature_columns = [
44      tf.feature_column.numeric_column('x', shape=np.array(x_train).shape[1:])]
45  regressor = tf.estimator.DNNRegressor(
46      feature_columns=feature_columns, hidden_units=[10, 10])
47
48  # Train.
49  train_input_fn = tf.estimator.inputs.numpy_input_fn(
50      x={'x': x_train}, y=y_train, batch_size=1, num_epochs=None, shuffle=True)
51  regressor.train(input_fn=train_input_fn, steps=2000)
52
53  # Predict.
54  x_transformed = scaler.transform(x_test)
55  test_input_fn = tf.estimator.inputs.numpy_input_fn(
56      x={'x': x_transformed}, y=y_test, num_epochs=1, shuffle=False)
57  predictions = regressor.predict(input_fn=test_input_fn)
58  y_predicted = np.array(list(p['predictions'] for p in predictions))
59  y_predicted = y_predicted.reshape(np.array(y_test).shape)
60
61  # Score with sklearn.
62  score_sklearn = metrics.mean_squared_error(y_predicted, y_test)
63  print('MSE (sklearn): {0:f}'.format(score_sklearn))
64
65  # Score with tensorflow.
66  scores = regressor.evaluate(input_fn=test_input_fn)
67  print('MSE (tensorflow): {0:f}'.format(scores['average_loss']))
68
69
70if __name__ == '__main__':
71  tf.app.run()
72