I'm trying to predict a time series, let's say I have 3 features and a target variable. I used the standard approach when feature lags have the same length, for example, 10. Then the size of my batch will be (32, 10, 3). But looking at the autocorrelation graphs, I want to use a lag length of 10 for the first feature, 5 for the second, and 7 for the third. How can I construct a batch in this case?
My architecture looks like this:
inp = Input(shape=(10, 3, batch_size=None)
x = LSTM(128, return_sequences=True)(inp)
x = Flatten()(x)
x = Dropout(0.2)(x)
out = Dense(2, activation='linear')(x)