Instructor:  Paul J. Smith, Statistics Program

Office:  MTH 4404
Telephone:  (301) 405-5104
E-mail:  pjs@math.umd.edu

Textbook:  Venables, W. N. and Ripley, B. D.   Modern Applied Statistics with S-PLUS (3rd ed.).
                           New York: Springer-Verlag.

Prerequisites:  STAT 420 or STAT 700.

Statistical research and application has changed dramatically because of cheap and powerful computational and graphical tools.  This course presents modern methods of computational statistics and their application to both practical problems and research.  The techniques described in STAT 798C should be part of every statistician's toolbox.

The prerequisite for this course is STAT 420 or STAT 700.  Statistical methodology will be presented informally, with emphasis on the intuitive basis for the technique and its implementation in the computing environment.  All methods will be illustrated by application to real-world data sets.

Topics:

• Introduction to S-Plus:  Starting and quitting S-Plus, on-line help, S-Plus operators and functions, creating S-Plus objects, data types (vectors, matrices, factors, frames, lists), managing data (combining objects, loops, subsetting, creation of frames), S-Plus graphics.
• Introduction to SAS:  The SAS environment, SAS data sets, sorting, combining and subsetting data, basic statistical procedures.
• Monte Carlo and Simulation in S-Plus:  Basic random number generation, applications of LLN and CLT in simulations, numerical integration, importance sampling, empirical distributions, Markov Chain Monte Carlo.
• Numerical Optimization in Statistics:  Objective functions in statistics, linear and nonlinear least squares, likelihood considerations, penalized likelihood, steepest descent, quasi-Newton-Raphson methods, constrained maximization, EM algorithm.
• Linear and Generalized Linear Models:  Regression summaries, fitting, prediction, model updating, analysis of residuals, model criticism, ANOVA, generalized linear models, specifying link and variance functions, stepwise model selection, deviance analysis.
•  Bootstrapping Methodology:  Parametric bootstrap, empirical CDF, bootstrap standard errors and confidence intervals, estimation of bias, jackknife, application to regression.
• Time Series:  Autocorrelation, spectral density and fast Fourier transform, window spectral estimates, ARMA modeling and forecasting, linear filtering.
• Smoothing and Nonparametric Regression:  Kernel density estimate, bandwidth selection, regression smoothing, simultaneous error bars, projection pursuit.

Grades will be based on computer assignments involving data analysis and statistical computation.

References:

 R. A. Becker, J. M. Chambers, and A. R. Wilks (1988).  The New S Language.  Pacific Grove, CA: Wadsworth & Brooks/Cole.

J.M. Chambers and T.J. Hastie (1993),  Statistical Models in S.  London: Chapman \& Hall.