Easy Last updated on April 26, 2022, 11:39 p.m.

`Answer In Progress`

The central limit theorem states that the expected value of a number of `(independent and identically distributed) i.i.d random variables`

with finite variances, will tend to a normal distribution as the number of random variables grows. More formally, if $X_{1}, X_{2}…$ are a sequence of i.i.d random variables with $E[X_{i}]=\mu$ and $Var[X_{i}]=\sigma^{2}<\infty$, then

$$\frac{\sqrt{n}\bar{X_{n}} - \mu}{\sigma} \rightarrow N(0,1)$$

A $\alpha$-confidence interval on the mean can then be computed as

$$[\bar{X_{n}} - z_{(1-\alpha)/2}\frac{\sigma}{n}, \bar{X_{n}} + z_{(1-\alpha)/2}\frac{\sigma}{n}]$$

Here $z$ denotes the z-score of the standard normal distribution. Note that we do not have access to $\sigma$ of the unknown distribution. In this case, we can use the standard deviation of the sample $\sigma_{est}$ and the z-score of the student t distribution to compute confidence intervals

$$[\bar{X_{n}} - z_{(1+\alpha)/2}\frac{\sigma_{est}}{\sqrt{n}}, \bar{X_{n}} + z_{(1+\alpha)/2}\frac{\sigma_{est}}{\sqrt{n}}]$$

where $z$ is the z-score of the student t distribution having degrees of freedom set to $n-1$

`Note:`

Bounds computed using the central limit theorem only holds for either a large number of samples or small samples sizes from Gaussian distributions. Sample sizes equal to or greater than 30 are often considered sufficient for the Central Limit Theorem to hold.

Frequently Asked Questions by

DeepMind