Exercises

These exercises ask you to assemble the habits from the Python subject into one small experiment.

Exercise: Workflow order

What is the workflow order used in this chapter?

Choose one

Select one choice, then check.

Hint

The workflow starts with configuration and ends with a saved or printed summary.

Solution

The workflow order is:

configuration -> data -> computation -> plot -> summary

That order makes the experiment easier to rerun and inspect.

Exercise: Make noisy data

Edit the code so it prints shapes: (4,) (4,).

Create paired arrays

Runs locally with Python in your browser.

Ready to run.

Hint

Use:

y = 2.0 * x + noise
Solution

One fix is:

The arrays match because noise was created with the same size as x.

Exercise: Compute a loss

Which expression computes mean squared error?

Answer it first, then check.

Hint

Mean squared error has three steps: subtract, square, average.

Solution

Use:

((prediction - y) ** 2).mean()

This computes squared errors and then averages them.

Exercise: Fit by search

Edit the code so it prints best: 2.0.

Choose the lowest-loss slope

Runs locally with Python in your browser.

Ready to run.

Hint

Inside the loop, compare loss with best_loss. If it is smaller, update both best_slope and best_loss.

Solution

One fix is:

The slope 2.0 gives zero loss for this exact data.

Exercise: Record the run

Name one field that should be saved in the summary so the experiment can be rerun.

Answer it first, then check.

Hint

The seed is the most direct answer because it controls the noisy data.

Solution

The most direct answer is:

seed

The summary should also record configuration values and results, but the seed is essential for recreating the same noisy data.