Scribd
Upload
89 views2 pages

Fundamentals of Statistics Guide

This document outlines the topics and schedule for the 18.650 Fundamentals of Statistics course taught by Roger Jin in Spring 2019. The course will cover introductions, Gaussian mixtures, maximum likelihood, sampling distributions, and will include an appendix with useful R commands. Key topics include modeling asset price evolution as a diffusion process, properties of sample means, and the mean squared error of an estimator.

Uploaded by

Roger Jin
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
89 views2 pages

Fundamentals of Statistics Guide

This document outlines the topics and schedule for the 18.650 Fundamentals of Statistics course taught by Roger Jin in Spring 2019. The course will cover introductions, Gaussian mixtures, maximum likelihood, sampling distributions, and will include an appendix with useful R commands. Key topics include modeling asset price evolution as a diffusion process, properties of sample means, and the mean squared error of an estimator.

Uploaded by

Roger Jin
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

18.

650 Fundamentals of Statistics


Roger Jin
Spring 2019

Contents
1 February 5, Introduction 1

2 February 12, Gaussian Mixtures 1

3 February 14, Maximum Likelihood 1


3.1 Diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

4 Sampling Distribution 2

5 Appendix 2
5.1 Useful R Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1 February 5, Introduction
1. We always call a parameter that we don’t know by a Greek letter, because we don’t know it.
2. No models are true, but some models are useful.

2 February 12, Gaussian Mixtures


Suppose that we have X ∼ N (θ1 , σ12 ) and Y ∼ N (θ2 , σ22 ). Then we can use linearity of expectation to
get the moments of a mixture of these two models.

3 February 14, Maximum Likelihood


3.1 Diffusion
We can model the evolution of the price of an asset as
dPt
= udt + σdWt ,
Pt
where W is a diffusion process. Then we can show that

Wt |W0 = 0 N (0, t).

Also apparently a Lalacian Distribution is a mixed Gaussian where the variances are themselves
exponentially distriuted.

1
4 Sampling Distribution
Distribution of the estimator, if the experiment were repeated √ many times.
Lecture 4 Slide 30 - without the multiplicative factor of n, the distribution converges to 0. Suppose
we have X1 , · · · , Xn that have mean θ and variance σ 2 . If we consider the sample mean
n
1X
X̄ = Xi
n i=1
σ2
Var X̄ =
n

Applying Chebyshev’s we have that

P X̄ − θ > k < k 2 Var X̄ = k 2 σ 2 /n,


 

√
which limits to 0. Then multiplying by n scales up the variance by n, so we no longer have this limit
go to zero. I still don’t get how it’s normal though, maybe just because of CLT.

Theorem 4.1 (MSE)

h i2
M SE(θ̂) = E θ̂ − θ
2
= Var θ̂ + (E [θ] − θ)

5 Appendix
5.1 Useful R Commands
1. str(A): displays structure of A .
2. ggplot(): plots stuff.

You might also like