Talk:SE250:lab-5:dols008

Comments from John Hamer

• In choosing the sample size, I'm assuming you performed some experiment. You should describe that experiment here.

• Nice table of results. A graph would be even nicer.

• delighted to have you record what you understand and don't understand.

• explain your choice of 60000/40000; it's an odd one.

• "Java string is debatably the best hash function" — explain the debate. The data you have is pretty clear, isn't it? But there's something uncomfortable about a function performing better than the theory predicts, especially as it wasn't so hot in the statistical tests.

• "Buzhash performed ... a fair bit worse for low entropy data" — really? Is 9 really worse than 8 when the expected is 8.8?

• "java object similarly terrible." — another surprise, and worth a discussion.

This is an impressive amount to show for two hours work. Admirably concise.