SFS 2025 Unlock the power of ecological insights in the era of big data! Massive ecological datasets are being generated more frequently and inexpensively than ever before. However, datasets that have far more predictors (P) than observations (N) pose serious challenges for traditional statistical methods, and often result in overfitting, poor predictive performance, and inaccurate variable selection. Sparse modeling approaches resolve the P>>N problem by constraining the number of potential predictors in a model, and typically lead to better out-of-sample prediction; these distilled models might also better reflect the true processes affecting the response. Join our interactive workshop where you'll delve into cutting-edge techniques using simple R packages to harness, analyze, and interpret big ecological datasets (datasets provided). This workshop is appropriate for scientists at any career stage; it will provide an introductory course on sparse modeling techniques followed by a live-coding lesson in R, where participants will create their own scripts for the SuSiE (susieR) and lasso (glmnet) approaches. Participants will walk away from this course with a firm understanding of several sparse modeling approaches and their usefulness, as well as methods for implementing, visualizing, and interpreting these modeling approaches in R.
Participants are asked to bring their own laptops.