STA 243: Computational Statistics

Subject: STA 243
Title: Computational Statistics
Units: 4.0
School: College of Letters and Science LS
Department: Statistics STA
Effective Term: 2009 Winter

Learning Activities

  • Lecture - 3.0 hours
  • Laboratory - 1.0 hours

Description

Numerical analysis; random number generation; computer experiments and resampling techniques (bootstrap, cross validation); numerical optimization; matrix decompositions and linear algebra computations; algorithms (markov chain monte carlo, expectation-maximization); algorithm design and efficiency; parallel and distributed computing.

Prerequisites

(STA 130A, STA 130B); (MAT 067 or MAT 167); Or equivalent of STA 130A and 130B, or equivalent of MAT 167 or MAT 067.

Expanded Course Description

Summary of Course Content: 
This course covers modern and classical methods of statistical computing and computational statistics. The course material includes algorithms, information theory, numerical computing, and computationally intensive methods (e.g. resampling, computer experiments). Emphasis will be put on algorithmic complexity, efficiency, optimization and data reduction. The goal is to give students a foundation in common computational techniques and algorithms (e.g. fast fourier transform, dynamic programming) that are used in the implementation of statistical methods (e.g. linear algebra calculations, matrix decompositions, random number generation). Examples will be drawn from diverse fields such as bioinformatics, ecology, medicine, computer vision, and stochastic finance. 
Students will use a general-purpose statistical programming language (e.g. R). We will discuss the use of other programming languages for various tasks (e.g. Matlab for large matrix calculations). Students will learn to design robust and accurate software to support their research. They will also to be able to critically understand the techniques used in existing software. Course work may involve complex computing tasks (e.g. interfacing between computer languages for more efficient code). 

Illustrative Reading: 
1) Computational Statistics by G Givens and J Hoeting. 
2) Exploratory Data Analysis with MATLAB by WL Martinez & AR Martinez.
3) Elements of Computational Statistics by JE Gentle.
4) Monte Carlo Strategies in Scientific Computing by J Liu. 

Potential Course Overlap: 
STA 242 is a graduate course with an emphasis on computing for statistical research. STA 243 covers numerical and computational issues in statistics, including algorithms and matrix calculations. The statistical programming taught in STA 243 involves implementing these computations and is totally complementary to that in STA 242. There is no overlap between STA 243 and STA 141.