Showing posts with label Theory. Show all posts
Showing posts with label Theory. Show all posts

Monday, June 4, 2012

Note on mathematical statistics: 6.2 Rao-Cramer lower bound and efficency

Mindmap for Section 6.2 Rao-Cramer lower bound and efficiency
Download MindJet file: https://docs.google.com/open?id=0Bw64rMSoJR_Za3FFMWM1LS1QZWs

Research Project on Sampling Rare Population

Term Research Project in Sampling Theory

Title: Sampling Rare Population


Outline
============================================
* Introduction
* Methods of SRP
Disproportional Stratified Sampling
Multiple Frames
Network (Multiplicity) Sampling
Snowball Sampling
Staged methods & Screening
* Example
* Research Note
* Concluding Remarks



Download slide in PDF: https://docs.google.com/open?id=0Bw64rMSoJR_ZM3hnRWNCMVN6THc
Download Reference: https://docs.google.com/document/d/1UsYYLU66HI0PBruoTE8tu6qbwFqVWtGJwdBe4C2OWB4/edit


Tuesday, May 22, 2012

Note on mathematical statistics: 5.7 Chi-Square Test

1. Mathematical basis of Chi-Square test
  • Multivariate normal distribution implies a Chi-Square (n) distribution of Y=Sum( (X_i-mu_i)^2 / sigma_i^2 ). 
  • By CLT, the joint pdf of different groups of sample are treated multivariate normal.
  • The statistic Q_k-1 = Sum( Y_i^2) = Sum [1..k] ( (X_i-E_i)^2/E_i ) has a Chi-Square(k-1) distribution.
  • With the idea of interval  frequency approximation, every distribution can be treated as Multi-binomial.
2. Procedure (the idea of interval  frequency approximation)
  • Partition the domain of experiment result into finite mutually disjoint sets A_1, A_2 ... A_n
  • Count the number of result in A_i as frequency X_i
  • Assign df  (different ways)
  • Assign the probability of result in A_i as p_i (different ways)
  • Evaluate statistic Q_k-1
  • Test
3. Three tests

Test Goodness of fit Homogeneity Independence
example 5.7.1, 5.7.2 5.7.3 5.7.4
H0 Result has the theoretical distribution Two sets of sample have the same distribution Two attributions of subjects are independent 
Key fact X_i/n = p_0i p_1i=p_2i=p_0i=E_i/nP_ij=Pi*Pj
source of E_i   multinomial modelMLE MLE
formula of E_i (1<=i<=k) E_i=n*p_0iE_i=(X_i1+X_i2)/(n1+n2) E_ij=(X_i./n) * (X_.j/n)
df k-1 k-1=2(k-1)-(k-1) (a-1)(b-1)=(a*b-1)-(a+b-2)
statistic Sum [i=1..k] ( (X_i-E_i)^2/ E_i ) Sum [j=1,2] [i=1..k]( (X_ij-n_j*E_i)^2/n_j*E_i ) Sum [j=1,a] [i=1..b]( (X_ij-n*E_ij)^2/n*E_ij)
Dataset x_1, x_2 ... x_n x_1, x_2,... x_n1
y_1, y_2 ... y_n2
contingency table
Xij, 1<=i<=a, 1<=j<=b
k=a*b

4. Remarks
  • Chi-Square tests are not exact test, but approximate test
  • The statistic is based on frequency of result in interval, instead of the result itself
  • Make sure E_i > 5 or use Fisher exact test
  • Minimum Chi-Square estimation
  • MLE based Chi-square tests have greater rejection rate than tests based on Minimum Chi-Square estimator
  • Every estimated parameter p0i costs one df

Notes on mathematical statistics: textbook

Here is a serial of notes on my studying on mathematical statistics, specifically on the textbook of
Introduction to Mathematical Statistics,
R.V. Hogg, A Craig and J. W. McKean
6th edition, Pearson.


There was an official solution book pdf file including partial answers to exercises.
However, it can be located and downloaded only after intensive Google search.

I keep my homework and exercises from Chapter 4 to Chapter 7 which can be used as an "AS IS" manual for the book (not only even number questions in official solution, but also some odd numbers questions).

Contact me with the page number and  exercises  index.
I will see what I can do for you.