CODEWITHSUNDEEP
python-3.xtensorflowkerasneural-networkrelu
325
325

Reputation: 606

Neural Network Using ReLU Activation Function

I am trying to use a neural network to predict the price of houses. Here is what the top of the dataset looks like:

    Price   Beds    SqFt    Built   Garage  FullBaths   HalfBaths   LotSqFt
    485000  3       2336    2004    2       2.0          1.0        2178.0
    430000  4       2106    2005    2       2.0          1.0        2178.0
    445000  3       1410    1999    1       2.0          0.0        3049.0

...

I am trying to use the ReLU activation function, but my accuracy is zero even after 100 epochs. Am I missing something here?

X = dataset[:,1:8] #predictor variables
Y = dataset[:,0] #sell price

#Normalize data
from sklearn import preprocessing
X_scale = min_max_scaler.fit_transform(X)
X_scale

#Split Data
from sklearn.model_selection import train_test_split
X_train, X_val_and_test, Y_train, Y_val_and_test = train_test_split(X_scale, Y, test_size=0.3)
X_val, X_test, Y_val, Y_test = train_test_split(X_val_and_test, Y_val_and_test, test_size=0.5)
print(X_train.shape, X_val.shape, X_test.shape, Y_train.shape, Y_val.shape, Y_test.shape)

from keras.models import Sequential
from keras.layers import Dense

model = Sequential(
    Dense(32, activation='relu', input_shape=(7,)))

model.compile(optimizer='sgd',
              loss='binary_crossentropy',
              metrics=['accuracy'])

hist = model.fit(X_train, Y_train,
          batch_size=32, epochs=100,
          validation_data=(X_val, Y_val))

model.evaluate(X_test, Y_test)[1]
## Output: 3/3 [==============================] - 0s 3ms/step - loss: -5698781.5000 - accuracy: 0.0000e+00

Upvotes: 0

Views: 207

Answers (2)

Jonathan Lorence
Jonathan Lorence

Reputation: 83

As well you are solving a regresion problem,

so you should use mean square as loss function...

and also you are trying to predict one values, so you should add one more layer to output that value.

Upvotes: 1

Timbus Calin
Timbus Calin

Reputation: 15063

Your accuracy is 0 because you forgot to add an output layer, so your loss is not computed properly. In addition to this, accuracy is not a relevant metric since you are doing regression and not classification.

You need to modify your model like this:

model = Sequential(
    Dense(32, activation='relu', input_shape=(7,)),
    Dense(1, activation='linear'))

Also, in your model.compile() you have to modify your loss to be "mse" instead of "binary_crossentropy", since you are doing regression and not classification.

model.compile(optimizer='sgd',
              loss='mse',
              metrics=['mean_squared_error'])

Upvotes: 1

Related Questions