Skip to content

Latest commit

 

History

History
30 lines (25 loc) · 1.67 KB

File metadata and controls

30 lines (25 loc) · 1.67 KB
layout default
title To Do

2025

Fix errors

  • In Common steps tutorial: PCA scale: needs to use scaling data (not at the moment)
  • In Common steps tutorial: need change running of RF, now it’s based on test data!
  • In Common steps tutorial or bagging / boosting / stacking: number of trees in RF is not tunned, add that (check also XGBoost)

For next time

  • Be clear about the course being mostly for omics analyses
  • Add basic ML to course prerequisites
  • Prepare primer on regularized methods, KNN, Random Forest, data splitting and ask people to study before the course
  • Prepare quiz checking understadning based on the primer, and even better, the entire Introduction to Biostatistics course
  • Common steps: add example of questions and answers around linear and logistics regression outputs
  • Common steps: add equations to all model evaluation metrics
  • Bagging, boosting, stacking lab: visualize XGBoost variable importance
  • Have Python code for all the labs
  • All chapters: check for missing learning outcomes, and preface text
  • Survival presentation: explain more time varying and competing risk scores
  • Survival lab: time varying a bit confusing example with start and stop intervals, maybe add one more example or replace this one
  • Put mix-effect models in one session, check lab, and more examples if needed
  • Have Introduction to Neural session in the Thursday afternoon
  • On Friday, continue with NN session: ML in LS applications, with additional focus on NN architecture, and more labs with image analyses, transfer learning etc. (see what people are doing and adjust)
  • Keep working in groups, in the morning let students discuss alone in the groups first!