Last class: Lesson 13: Model selection
- We discussed why model selection is important
- We cannot maximize the number of variables in the model
- We need to balance the bias-variance trade off
- We discussed how motivation for our research will guide our model selection technique
- Association vs. prediction
- Association: want to limit the number of covariates so we can get a more precise estimate (less variamce)
- Prediction: want to create a model that will help us with an accurate prediction of our outcome (less bias)
- We discussed several model selection techniques
- Focus on purposeful selection today
Announcements
- SPH Annual conference!
- I’m really behind on the muddy points!
- I have an hour today to work on them
- Reviewed the midterm feedback
- Thanks for your kinda words and constructive criticism
- I know this was really delayed - won’t see changes this quarter
- Makes me happy that many of you are benefitting from the flexible design of course
- Asking questions over Echo360
- I am sorry about this - running the class through Echo360 is really helpful to minimize the administrative work I need to do with Webex
- Also helps prevent my own human mistakes - no more forgetting to record
- Poster instructions are ready to go!
- There are examples (not exactly following our guidelines)
- There is a template for PPT and QMD
- On 3/17, we will have a poster day!!
- I will order some pizza!
- Feel free to bring extra snacks!
- You will need to informally print your poster (see instructions)
- Printed poster does not necessarily need to be what you turn in that night
- Meant to be a relaxed environment - no grading attached to it other than attendance (unless you spoke to me about previous commitments)
- You’ll be able to see others’ work and discuss different research questions!
Key Dates
- HW 4 due 2/28
- Lab 4 due 3/7
- HW 5 due 3/14
- Project poster due 3/17