Hydrologie – Environmental Statistics

Durchführung: Professur für Biometrie und Umweltsystemanalyse

1. repetition LM with multiple predictors: estimates, SE, P-values, interactions, non-linearity; plotting
2. count GLM: link function, Poisson distribution, back-transform to response scale; plotting
3. Bernoulli GLM: link function, Bernoulli distribution; backtransform; SE of prediction; plotting
4. other GLMs: canonical link function, neg binomial, binomial, any other: maximum log-likelihood principle; likelihood profile via grid search
5. Model diagnostics using simulation
6. missing data imputation (mice; sklearn.KNNImputer)
7. time series analysis with GAM and AR1
8. bootstrapping and jackknife
9. cross validation, incl. block cross-validation for time and space
10. randomForest
11. working with black-box models (ANN, SVM, …)
12. reproducibility

• quarto/markdown vs knitr/LaTeX
• meta-data for dryad and pangea
• docker, R: checkpoint for reproducibility

Lehrformen

• Vorlesungen
• Übungen
• Gruppenarbeit