Advanced Spatial Analysis and Statistics:
The following computer exercises are aimed at familiarizing each of you to the varying pieces of software and the statistical models used for developing species distribution models. As always, there are more than one way in which to complete an exercise. However, my designed workflow of each lab is how I feel best describes the concepts and processes. Enjoy...
Lab 1: An Introduction to Statistical Distributions
This lab is designed to help you better understand some of the different statistical distributions (families) and to also introduce you to Program R and Tinn-R (the R code editor).
Click here for Lab 1.
Lab 2: Random Sample and Pseudo-Absence Selection
As we have learned in lecture, selecting the most appropriate occurrence data is extremely important in the development of accurate SDMs. This lab is designed to provide you with an understanding of the concepts and a set of tools to assist with the selection of sample locations and the creation of pseudo-absences for creating an SDM.
Click here for Lab 2
Lab 3: Model Selection Using Information Theoretic Approach
The ability to develop multiple competing models of an ecological process and objectively determine the most appropriate model(s) is essential to making as informed of a decision as possible. We will explore the concepts and use of Information Theory for model selection through Akaike's Information Criterion (AIC).
Click here for Lab 3
Lab 4: Linear Regression Diagnostics - A Review
The content of this lab is to provide you with some of the necessary knowledge and tools to properly assess whether your model is an appropriate fit to the data. Ordinary Least Squares (OLS) regression is the most commonly applied modeling tool. Meeting the 4 basic assumptions of a linear model is paramount. The content of this lab provides a path to OLS success.
Click here for Lab 4
Lab 5: Spatial Analyses - Landscape Metrics and Attribution
In this lab we explore some of the simple methods commonly used to develop broad spatial scale metrics that we can then use as predictor variables in species distribution models. We will derive several different raster datasets then use Program R and the 'raster' package to attribute each sample location with the values from each raster.
Lab 6: Logistic Regression
Click here for Lab 6
Lab 7: Random Forests