ตัวเลขสุ่มและเมล็ดสุ่ม

Real datasets are not always handy. Random numbers let you build test data, simulate scenarios, and shuffle samples on demand.

Start by creating a generator with default_rng. This is the current, recommended way to make random numbers in NumPy.

Call random on the generator for floats between 0 and 1. Pass a size to get a whole array at once.

Use integers to draw whole numbers in a range. The low bound is included and the high bound is excluded.

The normal method samples a bell curve. Give it a mean and a standard deviation to shape the spread of values.

Run random code twice and you get different numbers. That breaks reproducibility when a teammate cannot recreate your results.

A seed sets the generator to a known start. Same seed means the exact same numbers, every single run.

Two generators built with seed 42 produce identical arrays. That is how notebooks stay reproducible across machines.

Generators also shuffle arrays and pick random samples with choice, perfect for splitting data or drawing test cases.

Seeding makes experiments repeatable. Reviewers can rerun your code and confirm they see the very same outcome you reported.

You may still see the older np.random.seed style. New code should prefer the default_rng generator for cleaner, safer randomness.

A colleague reruns your notebook and gets different random results.

You generated floats, ints, and normal samples, then locked them with a seed so your results are reproducible anywhere. 🎲