Course Description

In this course we will review many basic statistical methods you should be familiar with and apply them to various spatial datasets. In addition you will learn about several spatial statistical methods that you are probably not very familiar with and also apply them to several spatial datasets. Traditional statistical methods covered will be things like Ordinary Least Sqares Linear Regression (OLS), Analysis of Variance (ANOVA), characterizing uni-variate and multi-variate distributions, tests for normality (these usually fail with geographic data), Chi-square goodness of fit tests, t-tests and other difference of means tests. Other traditional statistical methods we will touch upon that are not necessarily spatial (but can be) are Cluster analysis, principle components analysis (PCA) and related Factor Analysis. The specifically spatial statistical methods we will cover are: 1) defining complete spatial randomness (CSR) for several different data models; 2) characterizing spatial autocorrelation with correlograms, variograms, and semi-variograms; 3) Various techniques of spatial interpolation with a focus on Inverse Distance Weighting (IDW) and Kriging; 4) Evaluating the skill or accuracy of spatial interpolations (kriging does this automatically but can be very wrong if assumptions of stationarity etc. do not hold) using a method known as cross-validation (aka ‘Jack-Knifing’, or ‘Bootstrapping’); 5) Spatial Modelling and Spatial Regression; and 6) Explore some of the problems with spatial statistics such as the Modifiable Areal Unit Problem (MAUP) and the Ecological Fallacy.

This course meets one evening a week and a typical class session will begin with two student powerpointe presentations reviewing each of last week’s GIA readings (~30 minutes); a new lecture from me on this week’s reading and some statistics (30-60) minutes. On weeks #3, #6, and #9 we will have teams present lab results for some of the lab time. Typically we will then take a short break and re-convene to immediately apply the methods recently discussed to spatial data in the lab using ArcView, ArcGIS, and a statistics package called JMP. This ‘computer time’ will be sufficient only to familiarize you with the techniques you will need to complete the analysis of the spatial dataset, NOT sufficient to complete the lab. In this course you will review and apply traditional statistical methods to spatial data using GIS and statistics software. In addition you will learn some more sophisticated tradtional statistical methods in addtion to some truly spatial statistical methods and apply them to spatial data using GIS and a statistics package. In addition you will be responsible for turning in four lab assignments which directly answer questions I have set for you; and, writing up with your group a ‘Betty Crocker’ lab manual for one of the spatial datasets complete with computer instructions, questions, and answers to questions. Your group will be responsible for 20% of the questions on the final that are directly related to the analysis of the dataset you wrote up an explicit lab for. Each group will contribute 20% of the points on the test and I will contribute the last 20%.

Texbooks and Software:

Geographic Information Analysis by Unwin and Sullivan (required)

JMP IN statistics package (student version) ~$60 (highly reccomended)

ArcView and Arc/INFO (we’ll use lab software for this)

CD of Datasets and course documents provided by Instructor ($10)

Other textbooks of interest that you may find useful:

Geostatistics

(industry standard on kriging, dull read but better than most others including Cressie)

Applied Statistics for Psychology by David Howell

(Great general statistics book)

The Cartoon Guide to Statistics by Matt Gronick

(great intro to statistics that steps you through basic conceptual stuff really well)

Grading

Lab Exercise #1: Spatial analysis of population survey in Santa Barbara County 10%

Due: In class Week #3

Lab Exercise #2: Correlation, Regression and Modeling Pop Density with DMSP imagery 10%

Due: In class Week #6

Lab Exercise #3: Spatial Interpolation: IDW & Kriging of a DEM, Rainfall data, & Ozone 10%

Due: In class Week #9

Group Exercise I: Complete ‘Betty Crocker’ write-up of one of Labs 1-3 15%

Group Exercise II: Powerpoint presentation of results of lab work. 10%

Powerpoint Summary of a Chapter in GIA 20%

Exam: (25% of questions from each group (3*25% = 75%), 25% from me ) 25%

Typical class sessions with go as follows:

First 30 minutes Two Powerpoint Review Sessions by Students

Next 30 minutes Overview of next 2 chapters in GIA by me

Next 30 minutes Statistical Lesson by me

Break

Computer Time for Labs and/or Lab Presentation weeks 3, 6, and 9

Course Schedule and Reading Assignments

By Week #2 (~March 30^th) you should have read Ch 1 and Ch 2

By Week #3 (~April 6^th) you should have read Ch 3 and Ch 4

By Week #4 (~April 13^th) you should have read Ch 5 and Ch 6

By Week #5 (~April 20^th) you should have read Ch 7 and Ch 8

By Week #6 (~April 27^th) you should have read Ch 9 and Ch 10

By Week #7 (~May 4^th) you should have read Ch 11 and Ch 12

By Week #8 (~May 11^th) provide questions for the exam and review

Week #9 EXAM (May 18^th)

Week #10 Wrap up loose ends and course evals