Рет қаралды 14,777
You might have learned that in some instances you don't divide by n when you calculate the empirical variance of your data but by n-1. This is known as Bessel's correction. In this video, I will explain when to use it, why we need a correction in the first place, and why this correction happens to be this weird n-1 thing you all know and wonder about.
0:00 - Intro
1:33 - What is variance?
4:28 - When to use the correction and what happens if you don't
9:20 - Explanation of bias involving sample mean
17:03 - Proof why n-1 eliminates bias involving sample mean
21:56 - Explanation and proof using pairwise differences
28:42 - Summary
If you want to read more about the topic:
en.wikipedia.org/wiki/Bessel'...
aaronjfisher.github.io/why-di...
stats.stackexchange.com/quest...
stats.stackexchange.com/quest...
www.quora.com/If-we-used-the-...