Stat 440: Statistical Computing

 

MWF:

Professor: Naomi Altman  http://www.stat.psu.edu/people/faculty/altman.html

 

prerequisites: some stat theory and methods

 

text:  Statistical Computing with R, Maria Rizzo, Chapman and Hall/CRC Press, 2008

 

Date

Lecture

Topic

Chapter

1/12

1

R Intro

1.1-1.4

1/14

2

Data types, reading and writing

1.6,1.7

R Doc

1/16

3

Scripts and Functions

1.5, 1.8

1/19

4

Simple Graphs

R Doc

1/21

5

Generating Random Numbers

3.1

1/23

6

Random Uniforms,

Seeds

3.2

1/26

7

More about random uniform

3.3

1/28

8

Inverse Integral

3.2,

1/30

9

Transform, Transformation etc

3.4,3.5

1/30

10

Accept/Reject Method

3.3

2/2-2/6

9

Mixture Distributions

Marron and Wand distributions,

etc

3.5

2/9

10

Matrix Algebra

properties of matrices

orthogonal matrices

symmetric matrices

Correlation and Covariance, Multivariate Normal

 

2/11

11

SVD and spectral decomp + applications

 

2/13

12

LU, QR and Choleski

 

2/16

13

CLT and the simulation study in HW

Expectation, Variance

When does the CLT hold.  How can you see if the CLT holds?

Is there a CLT for sample variance?

 

2/18

12

Expectation, Variance

2.1

2/20

18

Estimators

2.6,6.2

2/23

19

Bias, Variance, MSE

6.2

2/25

20

Intervals

2.6, 6.2

2/27

21

Tests

2.6, 6.3

3/2

22

Power

6.3

3/4

23

Monte Carlo Integration

5.1,5.2

3/6

24

Variance Reduction

5.3

break

 

 

 

3/16

25

Antithetic Variables

5.4

3/18

26

Control Variates

5.5

3/20

27

Importance Sampling

5.6

3/23

28

Stratified Sampling

5.7

3/25

29

Permutation Tests

8

3/27

44

Density Estimation

10.1,10.2

3/30

30

bootstrap

7.1

4/1

31

parametric, nonparametric and smooth bootstrap

7.1

4/3

33

confidence intervals

percentiles and pivots

 

4/6

34

more CIs

 

4/8

35

Bootstrap CIs - 5 types

7.4-7.5

4/10

36

Bootstrap in R

balanced bootstrap

7.4-7.5

4/13

37

Markov Chains

correlation (delta formula)

2.8

4/15

38

MCMC

9.1

4/17

39

Metropolis-Hastings

9.2

4/20

40

Gibbs

9.3

4/22

41

hybrid

9.2,9.3

4/24

42

convergence of algorithms

 

4/27

43

monitoring convergence

9.4

4/29

44

 finding zeroes

 

5/1

45

Newton Raphson

 

 

Grades

Homework:  50%

Project 1:      30%

Project 2:      20%

 

Letter Grades:

 

D      C      C+      B-      B       B+       A-       A

58     65     70       75      82      86       90       94

 

 

Homework

Homework will be assigned as needed.  There will be an ANGEL drop-box for each homework.  Turn in your R code and your answers.  The homeworks are each worth a different number of points, depending on the amount of work/learning involved.  For the final grade, the sum of your homework grades will be divided by 0.95*sum of homework totals.  The maximum homework score at the end of the semester is 40 points (no bonus points for obtaining more than 95% of the total grade).

 

You may collaborate on homework, but your final answer must be your own.

 

Project 1

I will pick a simulation problem.  Everyone in the class will do a different aspect of the simulations.  These will be presented in class before spring break.  All the results will be dropped into an ANGEL drop-box.  After break we will spend one class compiling the results into a paper or poster.  Your grade will depend on both the quality of your work and your presentation.  (There will be a grading rubric.)

 

Project 2

You will pick an individual topic.  Your code will be dropped into an ANGEL drop-box.  Your results will be reported as a presentation - PowerPoint or the online presentation format of your choice.  This is all due on the last day of classes.  We will use our 2-hour exam slot for presentations - probably 20 minutes each, depending on class size.  (There will be a grading rubric.)