MLP model 我做错了什么？

Question

I am developing a time series forecast model using python MLP model. My training sample has 550 events with 9 variables. I have a separate file for testing. I want to forecast at t+1 just one of the 9 variable. Since it is a time series I did 547 portions of 3 events walking one time each timefor X (547, 3, 9). And y (547,) is one variable at t+1.

model1 = Sequential()
model1.add(Dense(1, activation='tanh', input_shape=(3, 9)))
model1.add(Dense(1, activation='linear'))
model1.summary()

model1.compile(loss='mse', optimizer='adam', metrics=['mae'])

model1.fit(X, y, epochs=2000, verbose=1)

ValueError: Error when checking model input: the list of Numpy arrays that you are passing to your model is not the size the model expected. Expected to see 1 array(s), but instead got the following list of 547 arrays:

[array([[3.00000000e+01, 1.10000000e+01, 2.01500000e+03, 1.40000000e+01,
        8.36778275e-01, 2.07423694e-01, 7.65763170e-01, 5.07591141e-01,
        9.61668222e-01],
       [3.00000000e+01, 1.1000...

Answer 1

For a model with a single input you need to give a single array where the first dimension are the samples. You are giving a list. Convert to array (eg using np.stack(X)).