Applied Multivariate Data Analysis, UST 804
Fall Semester, 1999
Levin College of Urban Affairs
Cleveland State University
Instructor: Dr. William M. Bowen
Office UB107 Phone, 687-9226 Hours M – W 3 – 6 (or by appointment)

Course Description:

This course introduces the student to a range of multivariate methods used for analyzing large, complex and complicated data sets with a lot of variables, such as are used throughout empirical research on urban areas. It also includes a section on decision analysis.

The informal prerequisites for the course include competence at college algebra and elementary statistics, as well as ability to use a computer language or software package(s) with matrix algebra and the necessary multivariate statistics capabilities. The emphasis is on solving problems to obtain specific and accurate numerical answers.

Course Objectives:

  1. To introduce the student to the concepts behind why the following multivariate methods are and should be used in Urban Studies and/or Public Administration: principal components analysis, factor analysis, discriminant analysis, logistic regression analysis, cluster analysis, multivariate analysis of variance, multivariate generalizations of common univariate inference, and canonical correlation analysis.

    2. To introduce the student to how to use the preceding multivariate methods.

  2. To introduce the student to when or under what conditions to use each of the preceding multivariate methods.

  3. To introduce the student to what outcomes or products logically can or cannot be obtained when using each of these multivariate methods.

  4. To introduce the student to the method of mathematical programming and its extension to Data Envelopment Analysis.

  5. To introduce the student to methods for multi-criteria decision-making.

Text:

Johnson, Dallas E. (1998). Applied Multivariate Methods for Data Analysts. Pacific Grove, California: Duxbury Press.

Course Method:

Students are expected to prepare for and attend all classes, participate actively in discussions, and ask clarifying questions. Each week the student is expected to complete the week’s homework assignment, which will be given on Monday and due the next Monday. Also each week the student is expected to find a relevant research article using the method being examined that week, and prepare a 20 minute presentation on the article, with emphasis on the use of the method.

Grades:

Class participation: 20%

Homework: 80%

Tentative Schedule:

Week 1. Overview of Multivariate Data Analysis

 

Week 2. Sample Correlations

Week 3. Multivariate Data Plots

Week 4. Eigenvalues and Eigenvectors

Week 5. Principal Components Analysis

Week 6. Factor Analysis

Week 7. Discriminant Analysis

Week 8. Logistic Regression Methods

Week 9. Cluster Analysis

Week 10. Mean Vectors and Variance-Covariance Matrices

Week 11. Multivariate Analysis of Variance

Week 12. Factor Analysis and Regression

Week 13. Mathematical Programming

Week 14. Data Envelopment Analysis

Week 15. Decision Analysis