DLS 1 Week 3, Prediction for Exercise 9 - predict seems right, BUT

Hi folks,

it seems my prediction is right, but 1 out of three tests seemed to fail, see here

Do I miss something? Please let me know your thoughts and ideas

I tried an array around the array, but that does not help regarding the wrong shape error



You are using A2[0] for the comparison, so you are indexing off the first dimension. That gives an output that is a different shape than the input. The output will be a 1D vector, instead of a 2D vector.

It’s fine to do a Boolean comparison to a full array: it just gives you an output of the same shape with the elementwise comparison.


Dear Paul,

you made my day :slight_smile:

It works if I just write
Predictions = (A2 > 0.5)
Nevertheless, I tried the following as well

return [predictions]

This also created a 2d array, but seems not accepted by the autograder.

Thanks a lot



Hi, Frank.

I’m glad to hear that you found the solution. If by returning [predictions], you mean doing it your original way like this:

    predictions = (A2[0] > 0.5)
    print(f"predictions.shape = {predictions.shape}")
    return [predictions]

It turns out that doesn’t quite do the same thing as the correct solution that you showed. Here’s the result I get with the bracket solution:

parameters, t_X = predict_test_case()

predictions = predict(parameters, t_X)
print("Predictions: " + str(predictions))
print(f"predictions.shape = {predictions.shape}")


predictions.shape = (3,)
Predictions: [array([ True, False,  True])]
AttributeError                            Traceback (most recent call last)
<ipython-input-32-1a72e1f3efc9> in <module>
      3 predictions = predict(parameters, t_X)
      4 print("Predictions: " + str(predictions))
----> 5 print(f"predictions.shape = {predictions.shape}")
      7 predict_test(predict)

AttributeError: 'list' object has no attribute 'shape'

So you can see that the type of the result is still not correct.

You could make your 1D version work by doing this instead:

return np.reshape(predictions, (1,-1))

But we already have a simpler and cleaner solution, so what’s not to like about that?


1 Like