Tredennick model selection discussion questions

Questions

1. Why are the modeling goals of exploration, understanding (inference), and prediction incompatible? (In other words, why is it hard for one model to address more than one of these goals?)

2. On which of these three goals has your own research focused?

3. Why does linking ecological responses to weather represent such a difficult model selection problem? How do regularization methods used for predictive modeling, such as ridge regression and LASSO, “side-step” this problem?