Fair Shares in the Test Set

10 pts · 10 coins

Machine Learning in CodeEasy

Fair Shares in the Test Set

When you hold back test rows, you want every label to show up in the test set in the same share as in the whole dataset.

Task: write test_class_counts(name, test_size, seed). name is "iris", "wine" or "breast_cancer".

Load the dataset as X and y.
Split it with train_test_split(X, y, test_size=test_size, random_state=seed, stratify=y).
Return how many test rows have each label, as a list ordered by label number: [count of 0, count of 1, ...].

Examples

name = "iris"

test_size = 0.2

seed = 0

returns [10, 10, 10]

name = "wine"

test_size = 0.25

seed = 0

returns [15, 18, 12]

Hints

solution.py

1from sklearn.datasets import load_iris, load_wine, load_breast_cancer
2from sklearn.model_selection import train_test_split
3
4
5def test_class_counts(name, test_size, seed):
6    pass
7