Applied Multivariate Statistics for Ecological Data (ECO 632)

Lab Projects

The Applied Multivariate Statistics Lab includes several major projects corresponding to the major statistical techniques covered in this course. Each of these projects are briefly described below and include links to detailed descriptions of each assignment and downloads of all required software and data.

Biostats and Cartware R Packages

The biostats.R "package" is a compilation of functions written by Kevin McGarigal to support this and other offered statistics courses. This is not a real R package or library, but rather a compilation of R functions that must be sourced each time it is used. Similarly, Cartware.R is a suite of R functions written to facilitate classification and regression tree analysis. These pseudo-libraries are written in the spirit of growing up to become fully fledged R libraries, but for now they are just humble sets of functions that must be sourced before using.The biostats and Cartware functions are described fully in the accompanying help files.

Project 1 (Labs 1-2): Data screening and adjustment
This project is designed to give students hands-on experience using R to prepare, screen and potentially adjust a multivariate data set in preparation for analysis. Students will not be expected to evaluate the data set from every possible angle and using every possible analytical procedure available. Rather, the purpose of this project is to give students some basic experience in screening and adjusting data using R.
Project 2 (Labs 3-4): Finding groups -- cluster analysis

This project is designed to give students hands-on experience using Cluster Analysis (CA) to analyze a multivariate data set. As such, students will not be expected to analyze the data set from every possible angle and using every possible analytical procedure discussed in class. Rather, the purpose of this project is to give students some basic experience in how CA works and how R can be used to do the analysis.

Project 3 (Labs 5-8): Discrimination among groups

This project is designed to give students hands-on experience using a variety of procedures for testing, describing and predicting differences among groups using procedures such as Multi-response Permutation Procedures (MRPP), Analysis of Group Similarities (ANOSIM), Mantel Test (MANTEL), Discriminant Analysis (DA), and Classification and Regression Trees (CART). The purpose of this project is to expose students to these procedures for exploration, description and prediction of grouped data. As such, students will not be expected to analyze the data set from every possible angle and using every analytical procedure available. Rather, the purpose of this project is to give students some basic experience in how various discrimination techniques work and how R can be used to do the analysis.

Project 4 (Labs 9-11): Unconstrained ordination

This project is designed to give students hands-on experience using a variety of unconstrained ordination techniques, including Principal Components Analysis (PCA), Correspondence Analysis (CA), Detrended Correspondence Analysis (DCA), Principal Coordinates Analysis (PCO) and Nonmetric Multidimensional Scaling (NMDS) to analyze a multivariate data set. As such, students will not be expected to analyze the data set from every possible angle and using every possible analytical procedure available. Rather, the purpose of this project is to give students some basic experience in how unconstrained ordination techniques work and how R an be used to do the analysis.

Project 5 (Labs 12-13): Constrained Ordination

This project is designed to give students hands-on experience using constrained ordination (CCA/RDA/CAP) to analyze a multivariate data sets. Specifically, students will learn how to conduct Canonical Correspondence Analysis (CCA) or Redundancy Analysis (RDA) using R. Constrained Analysis of Principal Coordinates (CAP) is similar but will not be done here. Students will learn how to screen and adjust the data set (e.g., select species and explanatory variables) and conduct the constrained ordination. In addition, students will learn how to evaluate the performance of the model and interpret the relationships among species, samples and environmental variables. Finally, students will learn how to do a simple 2- or 3-part variance decomposition. Students will not be expected to analyze the data set from every possible angle and using every possible analytical procedure available. Rather, the purpose of this project is to give students some basic experience in how CCA/RDA work and how R can be used to do the analysis.