Statistical Methods for the Physical Sciences
Statistics M252
Atmospheric Sciences CM213

Professor: Robert Gould
Office: MS 6151
Hours: Wednesdays at 2pm & Fridays at 10am.
Phone: 20(6-3381)
email: rgould@stat.ucla.edu or rgould@ucla.edu



This course will give an introduction to basic statistical thought and practice with applications geared towards the physical sciences in general and atmospheric sciences in particular.  The emphasis will be on data analysis, and not mathematical theory.  However, out of necessity we will spend some time on the theory and review some mathematical and computational skills.

Getting Started:

1) Buy the book.
    You have a choice, depending on your discipline:

  Multivariate Geostatistics, Hans Wackernagel, Springer Verlag
 Astrostatistics, G.J. Babu and E.D. Feigelson, Chapman & Hall
 Statistical Methods in the Atmospheric Sciences, Daniel S. Wilks, Academic Press
2) Download the software
Go to http://forrest.psych.unc.edu/research/vista.html and download ViSta and the User's Manual.
3) You might also want to download and print (its in postscript format) a tutoral to Xlispstat called "A Surfer's guide to Lisp-Stat".  Chapter 2 is relevant to this course.

About the Computer:

We'll be meeting once a week (during class) in the Statistics computer lab on the 9th floor of Boelter Hall.  The computers are Macs (iMacs, actually) which I know causes untold consternation in some quarters.  Do not worry; you are allowed to use any computer you want or have access to.  The purpose of meeting in the lab is merely to demonstrate some statistical techniques, which you can then take with you.  However, you will need regularly access to a computer in order to complete the homework and exams.  So please see me immediately if you do not have access to a computer.

About the Software:
ViSta is a "front end" to the statistical programming language xlispstat.  It runs on Windows, Mac, and Unix systems.  ViSta is designed to analyze multi-variate data, and comes with much excess baggage.  Also, if you do not have multi-variate data, then it is a little like swatting a fly with a bomb.  We will therefore use xlispstat fairly frequently for more "simple" data sets.  The ViSta user's manual is a good introduction to that tool, and the "Surfer's Guide" gives you some idea of how to get started using xlisp-stat.  You are welcome to use any statistical package you are comfortable with; ViSta is recommended only because it is free and has good tools built-in.  The purpose of the course is to teach data analysis, not a software package.  So don't get caught up in the details of how to use the software.

Grading

Grades will be based on
Midterm during Week 6 (20%)  (See the midterm: requires adobe acrobat reader.)
Project (40%)
Final (40%)

Undergraduates enrolled can get credit for their homework (10%) and the final is then worth 30%.
Grades will be maintained on my.gradebook, which I understand is accessible via my.ucla.edu.
 
 

Homework
A list of homework problems.

Handouts
Basic Bootstrapping: notes from class.
Basic Bootstrapping lisp function.   Cut and paste it into the listener window to do very basic, somewhat clumsy, bootstrapping.
First xlisp-session.  From lecture notes. Requires Adobe Acrobat Reader.
Comparing estimators.  From lecture notes.  Discussion about median vs. mean with simulation results.
 

Data Sets
Most of the data sets used in this class are accessible bia this link.  This directory contains multiple versions of the data. This almost always includes the raw version, as well as a lisp version, and sometimes a Stata version.
 

Events of Interest
I'll try hard to list "extra-curricular" activities that might be of interest to students in this class (and others).