C4W4 Assignment: ERROR:: Incompatible shapes: [1150,64] vs. [1150] [Op:SquaredDifference]

Error :::

There was a problem grading your submission. Details:
Incompatible shapes: [1150,64] vs. [1150] [Op:SquaredDifference]

I thing Slicing pattern is not given Correctly. Can you say whether the slicing pattern is given correctly?

Compute the forecast for all the series

rnn_forecast = model_forecast(model, G.SERIES, G.WINDOW_SIZE).squeeze()

Slice the forecast to get only the predictions for the validation set

I think here is the problem…
rnn_forecast = rnn_forecast[G.SPLIT_TIME - G.WINDOW_SIZE:-1]

Please help me. I am deal this for last 10 days.

Please read this topic on how to interpret output shapes in a time series problem and fix your model.

There’s nothing wrong with the starter code. The slicing is done correctly.
Please use the model_forecast code as reference and fix your model.
print(model.summary()) if it helps.

Hi, I also facing this issues, it show my model is compatible to windowed datasets but also facing the incompatible shapes, may I know how to fix it?

Did you read the link (in the 1st response to the original post) I posted on how to interpret shapes in a time series problem?

Ya, thank you I have already read the link (in the 1st response to the original post). But I don’t know which parameter should I change, may I know which parameter should I change to solve the problem? Sorry for any inconvenience caused. Thank you.

Pay attention to the return_sequences parameter of the final recurrent layer.

1 Like