Engenharia de atributos básica no Keras avaliações

Mojtaba G. · Revisado há almost 4 years

Finally got through the lab after a few attempts, the use of the min max function here, which relies on pandas functionality, is not intuitive. I would have preferred if we instead used a normalization layer which is native to tensorflow. Furthermore, setting random seeds so the work is reproducible/verifiable would be helpful in quantifying the model improvements and incorporating that in the notebook. The instructions for predictions on the test dataset weren't very clear to me. I think there's a missed opportunity here to do things like [(X, y)] = test_ds.take(1), model.predict(X), and likewise feature_layer(X) to help the student understand what's happening under the good. Thank you!

Pritam D. · Revisado há almost 4 years

Samuel D. · Revisado há almost 4 years

Adriana B. · Revisado há almost 4 years

Ritish A. · Revisado há almost 4 years

Anna Elisabetta Z. · Revisado há almost 4 years

Snehashis J. · Revisado há almost 4 years

Mohammad A. · Revisado há almost 4 years

Mark M. · Revisado há almost 4 years

Dany T. · Revisado há almost 4 years

Victor P. · Revisado há almost 4 years

Supratim B. · Revisado há almost 4 years

Shivam S. · Revisado há almost 4 years

Alexander K. · Revisado há almost 4 years

Alexander K. · Revisado há almost 4 years

The instructions around get_scal are not clear, possibly not correct. It says get_scal takes a list of features. In your solution, it takes only one feature. The minmax function hardcodes train[feature]. Given what's previously taught in the class, I was expecting the use of tf.keras.layers.Normalization for this normalization. I think the language/solution could be improved upon. I'm going to try and do this lab on my own to try and get to the end solution. Thank you!

Pritam D. · Revisado há almost 4 years

Ayan D. · Revisado há almost 4 years

Liu S. · Revisado há almost 4 years

GAYATHRI D. · Revisado há almost 4 years

The lab is just copy and paste from the video

Santiago C. · Revisado há almost 4 years

Hyungil K. · Revisado há almost 4 years

Shashi P. · Revisado há almost 4 years

Hyungil K. · Revisado há almost 4 years