Data science: A reproducible split

A documented split rule lets anyone rebuild exactly the same training and test sets. Send every k-th row to the test set.

What to build

A program that splits rows into training and test sets by position.

Requirements

  • Counting positions from 1, a row goes to the test set when its position is divisible by k, and to the training set otherwise.
  • Print train N: ... and test M: ..., each followed by its rows in their original order, separated by spaces.

Input

Line 1: k. Line 2: the rows, as words separated by spaces.

Output

Two lines.

Constraints

1 ≤ k ≤ 100; 1 to 1000 rows.

Example

Input:

5
r1 r2 r3 r4 r5 r6 r7 r8 r9 r10

Output:

train 8: r1 r2 r3 r4 r6 r7 r8 r9
test 2: r5 r10

Why it works

A rule based only on position gives the same split every time it runs, so results can be reproduced and compared.

Examples

Example 1 · Input
5
r1 r2 r3 r4 r5 r6 r7 r8 r9 r10
Expected output
train 8: r1 r2 r3 r4 r6 r7 r8 r9
test 2: r5 r10
Example 2 · Input
2
a b c
Expected output
train 2: a c
test 1: b

Prompts inside input() are ignored when checking, so input("n: ") is fine.

Hints

Hint 1

Use enumerate(items, 1) to count positions from 1.

Hint 2

print("test", f"{len(test)}:", *test) prints the count and the items.

Solution

Show one correct solution (try it yourself first)
k = int(input())
items = input().split()
train = [x for i, x in enumerate(items, 1) if i % k != 0]
test = [x for i, x in enumerate(items, 1) if i % k == 0]
print(f"train {len(train)}:", *train)
print(f"test {len(test)}:", *test)

Lesson: The data science workflow →