W3_A1_Ex-7 Performance on other dataset not working

Course Q&A Deep Learning Specialization Neural Networks and Deep Learning
1

Re: 7- Performance on other datasets. The section says "If you want, you can rerun the whole notebook (minus the dataset part) for each of the following datasets. "

How do I get the whole notebook to run with an alternative data set? The original assignment uses from planar_utils import plot_decision_boundary, sigmoid, **load_planar_dataset**, load_extra_datasets AND load_planar_dataset()to import the original assignment data but when I tried switching the data out with the examples such as


datasets = {"noisy_circles": noisy_circles,
            "noisy_moons": noisy_moons,
            "blobs": blobs,
            "gaussian_quantiles": gaussian_quantiles}
### START CODE HERE ### (choose your dataset)
dataset = "gaussian_quantiles"
### END CODE HERE ###

I am able to get the data set to plot under the Ex-7 (see below) cell but it doesn’t propagate throughout the rest of the assignment so that the function calls get updated with the new data.

image
image1398×664 35.4 KB

I also tried to run each cell below 2 - Load the Dataset but the data set seemed to remain unchanged.

2

The datasets are just arguments that are passed to nn_model, right? So you invoke nn_model with the new dataset. You don’t need to rerun anything but the training with the new inputs. Just copy the code they gave you in the “Tuning the hidden layer size” section just above:

parameters = nn_model(X, Y, n_h, num_iterations = 5000)

That runs the training with X and Y as the dataset inputs. Of course you many want to adjust the number of iterations.

3

Or you could just run that cell they give you (with your choice the dataset) to set X and Y and then go back and rerun the “Tuning the hidden layer size” section. That will try training with the new dataset on the various different hidden layer sizes.