Draw 10 records at random, putting each back after it's picked. Some show up more than once, others not at all, so each tree in bagging gets different data.
Draw 10 records from the dataset, with replacement. Watch which records get picked more than once, and which get skipped.
This time, 5of the 10 original records weren't picked at all, while others were drawn more than once. Sampling with replacement means every draw is independent, so this changes each time.
