Feeding input into keras Sequential model

Question

I'm working on time series prediction with Keras. I have 10 timesteps in my input data which is of the shape (2688, 10, 1) i.e train_x.shape and train_y.shape is (2688, 10, 1). I am getting the following error while I try to feed it into the model.

ValueError: Error when checking target: expected activation_1 to have 2 dimensions, but got array with shape (2688, 10, 1)

The input shape that I am giving to the first lstm layer: input_shape=(1, time_steps) **

Am i reshaping train_y properly?

**

    time_steps=10
    train_x = np.reshape(train_x, (train_x.shape[0], train_x.shape[1], 1))
    train_y = np.reshape(train_y, (train_y.shape[0], train_y.shape[1], 1))

    # lstm model
    model = Sequential()
    model.add(LSTM(128, input_shape=(time_steps, 1), return_sequences=True))
    model.add(LSTM(64))
    model.add(Dense(1))
    model.add(Activation('linear'))

    model.compile(loss='mse', optimizer='adam')
    history = model.fit(train_x, train_y, epochs=10, validation_data=(test_x, 
    test_y), batch_size=64, verbose=1)

Tried it did not work...doing that gives me the following error ValueError: Error when checking target: expected activation_1 to have 2 dimensions, but got array with shape (2688, 10, 1) — Arpith S
– Arpith S, Commented Mar 4, 2019 at 5:30
You should probably post your code then. I can't really give anything but wild guesses at this point. — Primusa
– Primusa, Commented Mar 4, 2019 at 5:30

Fasty · Accepted Answer · 2019-03-04 05:44:46Z

1

train_x = np.reshape(train_x, (train_x.shape[0], train_x.shape[1], 1))
train_y = np.reshape(train_y, (train_y.shape[0], train_y.shape[1], 1)

I think the error is here, you should have given the time steps between shape[0] and shape[1] i.e...

train_x = np.reshape(train_x, (train_x.shape[0], 10,train_x.shape[1]))
train_y = np.reshape(train_y, (train_y.shape[0],10, train_y.shape[1]))

Here the value '10' denotes the time steps!

answered Mar 4, 2019 at 5:44

Fasty

8141 gold badge14 silver badges38 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

Arpith S Over a year ago

Tried it...gives me this error: ValueError: cannot reshape array of size 26880 into shape (2688,10,10)

Fasty Over a year ago

yes when you give timesteps as 10, you should make sure the product should also add up to the same product! i.e.. suppose your original shape 2688 * 10, the reshaped matrix should be 2688*10*1. Your product turns up to 268800!! which is wrong it should only come to 26880. make your reshaped matrix such that it should come to product 26880

Arpith S Over a year ago

Yes that is what I was doing originally -> train_x = np.reshape(train_x, (train_x.shape[0], train_x.shape[1], 1)) . It still does not work

Hasitha Amarathunga · Accepted Answer · 2019-03-04 05:29:55Z

0

if you expected shape (2688, 10, 1) then it cant be input_shape=(1, time_steps). It should input_shape=(time_steps,1)

answered Mar 4, 2019 at 5:29

Hasitha Amarathunga

2,02114 silver badges19 bronze badges

5 Comments

Arpith S Over a year ago

Tried it did not work...doing that gives me the following error ValueError: Error when checking target: expected activation_1 to have 2 dimensions, but got array with shape (2688, 10, 1)

Hasitha Amarathunga Over a year ago

Is it your datasets images size is 2688x10 pixel?

Arpith S Over a year ago

No, it is stock prices...i have posted the code please take a look

Hasitha Amarathunga Over a year ago

Please change input_shape=( 1,time_steps) and try

Arpith S Over a year ago

Does not work...gives me this error ValueError: Error when checking input: expected cu_dnnlstm_1_input to have shape (1, 10) but got array with shape (10, 1)

Collectives™ on Stack Overflow

Feeding input into keras Sequential model

2 Answers 2

3 Comments

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related