Look Inside

A fitted pipeline isn't a black box. Each step is still there, carrying what it learned.

Task: write inspect_pipeline(train_rows, train_labels).

• train_rows is a list of rows of numbers, and train_labels the answers. There are at least 3 rows.
• Build make_pipeline(StandardScaler(), KNeighborsClassifier(n_neighbors=3)) and fit it on the rows.
• Return a dict:
  • "steps": the names of the pipeline's steps, in order, as a list of strings.
  • "means": the column averages the scaler learned, as a list, each rounded to 2 decimal places.