In [1]:

    

import keras 
import numpy as np


    

    




Using TensorFlow backend.

In [2]:

    

from keras.datasets import boston_housing, mnist , cifar10, imdb


    

In [4]:

    

(x_train, y_train), (x_test, y_test) = boston_housing.load_data()


    

    




Downloading data from https://s3.amazonaws.com/keras-datasets/boston_housing.npz
40960/57026 [====================>.........] - ETA: 0s

In [5]:

    

print(x_train.shape)
print(x_test.shape)


    

    




(404, 13)
(102, 13)

In [7]:

    

print(y_train.shape)
print(y_test.shape)


    

    




(404,)
(102,)

In [8]:

    

from keras.models import Sequential # model 이름 t순차적으로 들어 간다는 뜻
from keras.layers import Dense  # fully connected layer
from keras import losses
from sklearn.metrics import mean_squared_error


    

In [26]:

    

model = Sequential()
model.add(Dense(13, input_dim =x_train.shape[1], kernel_initializer = 'normal', activation ='relu'))
model.add(Dense(9,kernel_initializer = 'normal', activation ='relu'))
model.add(Dense(7,kernel_initializer = 'normal', activation ='relu'))
model.add(Dense(3,kernel_initializer = 'normal', activation ='relu'))
model.add(Dense(1,kernel_initializer = 'normal'))
#텐서플로는 bias등 여러가지 설정을 해주어야 하지만 keras는 그냥 모델만 설정해주면 간편히 만든다.


    

In [27]:

    

model.compile(loss = 'mean_squared_error', optimizer = 'adam')


    

In [28]:

    

model.fit(x_train, y_train, batch_size =30, epochs = 15, verbose =1)


    

    




Epoch 1/15
404/404 [==============================] - 0s - loss: 585.8495     
Epoch 2/15
404/404 [==============================] - 0s - loss: 584.9909     
Epoch 3/15
404/404 [==============================] - 0s - loss: 583.1758     
Epoch 4/15
404/404 [==============================] - 0s - loss: 577.0550     
Epoch 5/15
404/404 [==============================] - 0s - loss: 556.4622     
Epoch 6/15
404/404 [==============================] - 0s - loss: 496.2252     
Epoch 7/15
404/404 [==============================] - 0s - loss: 366.2123     
Epoch 8/15
404/404 [==============================] - 0s - loss: 182.4815     
Epoch 9/15
404/404 [==============================] - 0s - loss: 138.2685     
Epoch 10/15
404/404 [==============================] - 0s - loss: 127.6726     
Epoch 11/15
404/404 [==============================] - 0s - loss: 120.9002    
Epoch 12/15
404/404 [==============================] - 0s - loss: 115.9808     
Epoch 13/15
404/404 [==============================] - 0s - loss: 111.9896     
Epoch 14/15
404/404 [==============================] - 0s - loss: 107.8595     
Epoch 15/15
404/404 [==============================] - 0s - loss: 104.2278     






    
Out[28]:





<keras.callbacks.History at 0x1d184fb5ba8>

In [18]:

    

y_pred = model.predict(x_test)


    

In [19]:

    

mean_squared_error(y_pred, y_test)


    

    
Out[19]:





119.88628633176269

In [ ]: