Spatial Predictions and Model Assesment
Content for Monday, December 1, 2025
We’ve spent much of the last few weeks developing different statistical models for explaining spatial patterns. We’ve moved from naive models os spatial covariance (based on semi-variance), to regression models attempting to identify the “weights” (i.e., slopes or \(\beta\)s) associated with different spatial predictors, to spatial models that account for violations of regression assumptions due to autocorrelation. While we’ve explored how each of these models can be sensitive to different challenges with spatial data and how that can effect your interpretation of variables in the models, we haven’t said much about how to know whether or not your model is any good in the first place. We’ll do that today and begin to introduce spatial predictions along the way!
Resources
A practical guide to selecting models for exploration, inference, and prediction in ecology by (Tredennick et al. 2021) highlights the importance of understanding model performance before making inference on predictor effects.
Model selection using information criteria, but is the “best” model any good? by (Mac Nally et al. 2018) highlights the importance of understanding model performance before making inference on predictor effects.
Standards for distribution models in biodiversity assessments by (Araújo et al. 2019) highlights the importance of understanding model performance before making inference on predictor effects.
Objectives
By the end of today you should be able to:
Articulate three different reasons for modeling and how they link to assessments of fit
Describe and implement several test statistics for assessing model fit
Describe and implement several assessments of classification
Describe and implement resampling techniques to estimate predictive performance