Tree-Based Models Learn decision trees, random forests, and boosted models in R. This one-day live online course covers CART, bagging, boosting, and model interpretation for applied data analysis.
https://www.prstats.org/course/tree-based-models-tbmr01/ *Unlock the Power of Tree-Based Models: a One-Day Workshop in R* Complex relationships, nonlinear effects, and high-dimensional predictors are everywhere in applied research. *Tree-Based Models (TBMR01)* is a live, hands-on course that teaches decision trees, random forests, boosting, and model interpretation — all in R. ------------------------------ Why this course is valuable - Tree methods are *flexible and robust* — they make few distributional assumptions and can automatically handle complex interactions and nonlinearities. - Ensemble methods (bagging, random forests, boosting) dramatically improve predictive performance while controlling overfitting — skills useful in ecology, public health, environmental science, and data science generally. - You’ll gain tools not just to *predict*, but to *explain* your models — via variable importance, partial dependence, SHAP or PDP visualisations. - Because the course is outcome-oriented and uses real data, you’ll come away able to apply these methods immediately in your own work. ------------------------------ What you will learn - The logic of decision trees: recursive partitioning, splitting rules (Gini, deviance, variance) - How to avoid overfitting: tuning, pruning, cross-validation, complexity control - Ensemble methods: bagging and random forests — implementation, interpretation, out-of-bag error estimation - Boosting (e.g. *xgboost*, *lightgbm*): algorithmic principles, hyperparameter tuning, advanced interpretation tools - Using interpretability tools: variable importance metrics, partial dependence plots, SHAP values or related methods ------------------------------ Format & logistics - *Duration*: 1 day, 6 hours - *Next date*: December 2, 2025 - *Format*: Live online; sessions recorded for later access - *Fee*: £150 - *Intended audience*: Ecologists, environmental scientists, public health analysts, data scientists, postgraduate students, early-career researchers - *Prerequisites*: Basic experience with R/RStudio; familiarity with descriptive stats and linear regression will help - *Support & materials*: Full code, data, slides shared; 30 days post-course email support; participants encouraged to bring their own data ------------------------------ If you’re working with complex data and want to build models that are both powerful and interpretable, this course delivers. Email [email protected] with any questions -- Oliver Hooker PhD. PR stats [[alternative HTML version deleted]] _______________________________________________ R-sig-ecology mailing list [email protected] https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
